Megatest

Artifact [e0bbf25b11]
Login

Artifact e0bbf25b11d736d1e9ba19310a2efce909cae7b5:


;;======================================================================
;; Copyright 2017, Matthew Welland.
;; 
;; This file is part of Megatest.
;; 
;;     Megatest is free software: you can redistribute it and/or modify
;;     it under the terms of the GNU General Public License as published by
;;     the Free Software Foundation, either version 3 of the License, or
;;     (at your option) any later version.
;; 
;;     Megatest is distributed in the hope that it will be useful,
;;     but WITHOUT ANY WARRANTY; without even the implied warranty of
;;     MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
;;     GNU General Public License for more details.
;; 
;;     You should have received a copy of the GNU General Public License
;;     along with Megatest.  If not, see <http://www.gnu.org/licenses/>.

;;======================================================================

(declare (unit runsmod))

(declare (uses debugprint))
(declare (uses commonmod))
(declare (uses configfmod))
(declare (uses mtargs))
(declare (uses mtver))
(declare (uses csv-xml))
(declare (uses keysmod))
(declare (uses mtmod))
(declare (uses processmod))
(declare (uses dbmod))
(declare (uses rmtmod))
(declare (uses testsmod))
(declare (uses tasksmod))
(declare (uses archivemod))
(declare (uses launchmod))
(declare (uses subrunmod))
(declare (uses servermod))
(declare (uses itemsmod))

(module runsmod
	*
	
(import scheme
	(prefix sqlite3 sqlite3:)
	chicken.base
	chicken.condition
	chicken.eval
	chicken.file
	chicken.file.posix
	chicken.format
	chicken.io
	chicken.pathname
	chicken.port
	chicken.pretty-print
	chicken.process
	chicken.process-context
	chicken.process-context.posix
	chicken.sort
	chicken.string
	chicken.time
	chicken.time.posix
	chicken.random
	chicken.process.signal
	
	(prefix base64 base64:)
	csv-xml
	directory-utils
	format
	matchable
	regex
	s11n
	srfi-1
	srfi-13
	srfi-18
	srfi-69
	stack
	sxml-modifications
	system-information
	typed-records
	z3

	(prefix mtargs args:)
	commonmod
	configfmod
	debugprint
	keysmod
	mtmod
	mtver
	processmod
	dbmod
	rmtmod
	testsmod
	tasksmod
	archivemod
	launchmod
	subrunmod
	servermod
	itemsmod
	
	)

;; (include "db_records.scm")
(include "run_records.scm")
(include "test_records.scm")
(include "key_records.scm")

;; use this struct to facilitate refactoring
;;

(defstruct runs:dat
  reglen regfull
  runname max-concurrent-jobs run-id
  test-patts required-tests test-registry
  registry-mutex flags keyvals run-info all-tests-registry
  ;; stores results from last runs:can-run-more-tests 
  (can-run-more-tests      #f) ;; (list can-run-more-flag num-running num-running-in-jobgroup max-concurrent-jobs job-group-limit)
  ((can-run-more-tests-count 0) : fixnum)
  (last-fuel-check         0)  ;; time when we last checked fuel
  (beginning-of-time       (current-seconds))
  (load-mgmt-function      #f)
  (wait-for-jobs-function  #f)
  (last-load-check-time    0)
  (last-jobs-check-time    0)
  )

(defstruct runs:testdat
  hed tal reg reruns  test-record
  test-name item-path jobgroup
  waitons testmode  newtal itemmaps prereqs-not-met)
  
;; look in the $MT_RUN_AREA_HOME/.softlocks directory for key-host-pid.softlock files
;;  - remove any that are over 3600 seconds old
;;  - if there are any that are younger than 10 seconds
;;      * sleep 10 seconds
;;      * touch my key-host-pid.softlock file
;;      * return
;;  - if there are no files younger than 10 seconds
;;     * touch my key-host-pid.softlock file
;;     * return
;;  
(define (runs:wait-on-softlock rdat key)
  (if (not (and *toppath* (file-exists? *toppath*))) ;; don't seem to have toppath yet
      (debug:print-info 0 *default-log-port* "Can't create softlocks - don't see MTRAH yet.")
      (let* ((softlocks-dir (conc *toppath* "/.softlocks")))
	(if (not (file-exists? softlocks-dir))
	    (create-directory softlocks-dir #t))
	(let* ((my-lock-file   (conc softlocks-dir "/" key "-" (get-host-name) "-" (current-process-id) ".softlock"))
	       (lock-files     (filter (lambda (x)
					 (not (equal? x my-lock-file)))
				       (glob (conc softlocks-dir "/" key "*.softlock"))))
	       (fresh-locks    (any (lambda (x) ;; do we have any locks younger than 10 seconds
				      (let* ((mod-time (file-modification-time x))
					     (age      (- (current-seconds) mod-time)))
					(cond
					 ((> age 3600) ;; too old to keep, remove it
					  (delete-file* x) #f)
					 ((< age 10)  #t)
					 (else #f))))
				    lock-files)))
	  (if fresh-locks
	      (begin
		(if (runs:lownoise "runners-softlock-wait" 360)
		    (debug:print-info 0 *default-log-port* "Other runners in flight, giving up some time..."))
		(thread-sleep! 2))
	      (begin
		(if (runs:lownoise "runners-softlock-nowait" 360)
		    (debug:print-info 0 *default-log-port* "No runners in flight, updating softlock"))
		(let* ((ouf (open-output-file my-lock-file)))
		  (with-output-to-port ouf
		    (lambda ()(print (current-seconds))))
		  (close-output-port ouf))))
	  (runs:dat-last-fuel-check-set! rdat (current-seconds))))))
  
;; Fourth try, do accounting through time
;;
(define (runs:parallel-runners-mgmt rdat)
  (let ((time-to-check (configf:lookup-number *configdat* "runners" "time-to-check" default: 10)) ;; 28
	(time-to-wait  (configf:lookup-number *configdat* "runners" "time-to-wait" default: 30))
	(now-time      (current-seconds)))
    (if (> (- now-time (runs:dat-last-fuel-check rdat)) time-to-check) ;; time to check
	(runs:wait-on-softlock rdat "runners"))))

;; To test parallel-runners management start a repl:
;;  megatest -repl
;; then run:
;;  (runs:test-parallel-runners 60)
;;
(define (runs:test-parallel-runners duration #!optional (proc #f))
  (let* ((rdat   (make-runs:dat))
	 (rtime  0)
	 (startt (current-seconds))
	 (endt   (+ startt duration)))
    ((or proc runs:parallel-runners-mgmt) rdat)
    (let loop ()
      (let* ((wstart (current-seconds)))
	(if (< wstart endt)
	    (let* ((work-time (pseudo-random-integer 10)))
	      #;(debug:print-info 0 *default-log-port* "working for " work-time
				" seconds. Total work: " rtime ", elapsed time: " (- wstart startt))
	      (thread-sleep! work-time)
	      (set! rtime (+ rtime work-time))
	      ((or proc runs:parallel-runners-mgmt) rdat)
	      (loop)))))
    (let* ((done-time (current-seconds)))
      (debug:print-info 0 *default-log-port* "DONE: rtime=" rtime ", elapsed time=" (- done-time startt)
			", ratio=" (/ rtime (- done-time startt))))))

;; Every time can-run-more-tests is called increment the delay
;;
;; NOTE: We run this server-side!! Do not use this global except in the runs:can-run-more-tests routine
;;
;; (define *runs:can-run-more-tests-count* 0)
(define (runs:shrink-can-run-more-tests-count runsdat)
  (runs:dat-can-run-more-tests-count-set! runsdat 0))

(define (runs:inc-can-run-more-tests-count runsdat)
  (runs:dat-can-run-more-tests-count-set!
   runsdat
   (+ (runs:dat-can-run-more-tests-count runsdat) 1)))

;;  (set! *runs:can-run-more-tests-count* 0)) ;; (/ *runs:can-run-more-tests-count* 2)))

;; Temporary globals. Move these into the logic or into common
;;
(define (runs:inc-cant-run-tests testname)
  (hash-table-set! *seen-cant-run-tests* testname
		   (+ (hash-table-ref/default *seen-cant-run-tests* testname 0) 1)))

(define (runs:can-keep-running? testname n)
  (< (hash-table-ref/default *seen-cant-run-tests* testname 0) n))

;; mechanism to limit printing info to the screen that is repetitive.
;;
;; Example: 
;; (if (runs:lownoise "waiting on tasks" 60)
;;     (debug:print-info 2 *default-log-port* "waiting for tasks to complete, sleeping briefly ..."))
;;
(define (runs:lownoise key waitval)
  (let ((lasttime (hash-table-ref/default *runs:denoise* key 0))
	(currtime (current-seconds)))
    (if (> (- currtime lasttime) waitval)
	(begin
	  (hash-table-set! *runs:denoise* key currtime)
	  #t)
	#f)))

(define (runs:can-run-more-tests runsdat run-id jobgroup max-concurrent-jobs)

  ;; Take advantage of a good place to exit if running the one-pass methodology
  (if (and (> (runs:dat-can-run-more-tests-count runsdat) 20)
	   (args:get-arg "-one-pass"))
      (exit 0))

  (if (runs:dat-load-mgmt-function runsdat)((runs:dat-load-mgmt-function runsdat)))

  (let* ((num-running             (rmt:get-count-tests-running run-id))
	 (num-running-in-jobgroup (rmt:get-count-tests-running-in-jobgroup run-id jobgroup))
	 (job-group-limit         (let ((jobg-count (configf:lookup *configdat* "jobgroups" jobgroup)))
				    (if (string? jobg-count)
					(string->number jobg-count)
					jobg-count))))
    (if (> (+ num-running num-running-in-jobgroup) 0)
	(runs:inc-can-run-more-tests-count runsdat)) ;; (set! *runs:can-run-more-tests-count* (+ *runs:can-run-more-tests-count* 1)))
    (if (not (eq? *last-num-running-tests* num-running))
	(begin
	  (debug:print 2 *default-log-port* "max-concurrent-jobs: " max-concurrent-jobs ", num-running: " num-running)
	  (set! *last-num-running-tests* num-running)))
    (if (not (eq? 0 *globalexitstatus*))
	(list #f num-running num-running-in-jobgroup max-concurrent-jobs job-group-limit)
	(let* ((can-not-run-more (cond
				 ;; if max-concurrent-jobs is set and the number running is greater 
				 ;; than it then cannot run more jobs
				 ((and max-concurrent-jobs (>= num-running max-concurrent-jobs))
				  (if (runs:lownoise "mcj msg" 60)
				      (debug:print 0 *default-log-port* "WARNING: Max running jobs exceeded, current number running: " num-running 
						   ", max_concurrent_jobs: " max-concurrent-jobs))
				  #t)
				 ;; if job-group-limit is set and number of jobs in the group is greater
				 ;; than the limit then cannot run more jobs of this kind
				 ((and job-group-limit
				       (>= num-running-in-jobgroup job-group-limit))
				  (if (runs:lownoise (conc "maxjobgroup " jobgroup) 60)
				      (debug:print 1 *default-log-port* "WARNING: number of jobs " num-running-in-jobgroup 
						   " in jobgroup \"" jobgroup "\" exceeds limit of " job-group-limit))
				  #t)
				 (else #f))))
	  (list (not can-not-run-more) num-running num-running-in-jobgroup max-concurrent-jobs job-group-limit)))))

(define (runs:run-pre-hook run-id)
    (let* ((run-pre-hook   (configf:lookup *configdat* "runs" "pre-hook"))
           (existing-tests (if run-pre-hook
                               (rmt:get-tests-for-run run-id "%" '() '() ;; run-id testpatt states statuses
                                                      #f #f ;; offset limit
                                                      #f ;; not-in
                                                      #f ;; sort-by
                                                      #f ;; sort-order
                                                      #f ;; get full data (not 'shortlist)
                                                      0 ;; (runs:gendat-inc-results-last-update *runs:general-data*) ;; last update time
                                                      'dashboard)
                               '()))
           (log-dir         (conc *toppath* "/logs"))
           (log-file        (conc "pre-hook-" (string-translate (get-environment-variable "MT_TARGET") "/" "-") "-" (get-environment-variable "MT_RUNNAME") ".log"))
           (full-log-fname  (conc log-dir "/" log-file)))
      (if run-pre-hook
          (if (null? existing-tests)
              (let* ((use-log-dir (if (not (directory-exists? log-dir))
                                      (handle-exceptions
                                       exn
                                       (begin
                                         (debug:print 0 *default-log-port* "WARNING: Failed to create " log-dir ", exn=" exn)
                                         #f)
                                       (create-directory log-dir #t)
                                       #t)
                                      #t))
                     (start-time   (current-seconds))
                     (actual-logf  (if use-log-dir full-log-fname log-file)))
                (handle-exceptions
                 exn
                 (begin
                   (print-call-chain *default-log-port*)
                   (debug:print 0 *default-log-port* "Message: " ((condition-property-accessor 'exn 'message) exn) ", exn=" exn)
                   (debug:print 0 *default-log-port* "ERROR: failed to run pre-hook " run-pre-hook ", check the log " log-file))
                 (debug:print-info 0 *default-log-port* "running run-pre-hook: \"" run-pre-hook "\", log is " actual-logf)
                 (system (conc run-pre-hook " >> " actual-logf " 2>&1"))
                 (debug:print-info 0 *default-log-port* "pre-hook \"" run-pre-hook "\" took " (- (current-seconds) start-time) " seconds to run.")))
              (debug:print 0 *default-log-port* "Skipping pre-hook call \"" run-pre-hook "\" as there are existing tests for this run.")))))
   



;; return #t when all items in waitors-upon list are represented in test-patt, #f otherwise.
(define (runs:testpatts-mention-waitors-upon? test-patt waitors-upon)
  (null? (tests:filter-test-names-not-matched waitors-upon test-patt)))

;;======================================================================
;; runs:run-tests is called from megatest.scm and itself
;;======================================================================
;;
;;  test-names: Comma separated patterns same as test-patts but used in selection 
;;              of tests to run. The item portions are not respected.
;;              FIXME: error out if /patt specified
;;            
(define (runs:run-tests target runname test-patts user flags #!key (run-count 1)) ;; test-names
  (let* ((keys               (keys:config-get-fields *configdat*))
	 (keyvals            (keys:target->keyval keys target))
	 (run-id             (rmt:register-run keyvals runname "new" "n/a" user (args:get-arg "-contour")))  ;;  test-name)))
	 ;; (deferred          '()) ;; delay running these since they have a waiton clause
	 (runconfigf         (conc  *toppath* "/runconfigs.config"))
         (dbfile             (conc  *toppath* "/.db/main.db"))
         (readonly-mode      (not (file-writable? dbfile)))
	 (test-records       (make-hash-table))
	 ;; need to process runconfigs before generating these lists
	 (all-tests-registry #f)  ;; (tests:get-all)) ;; (tests:get-valid-tests (make-hash-table) test-search-path)) ;; all valid tests to check waiton names
	 (all-test-names     #f)  ;; (hash-table-keys all-tests-registry))
	 (test-names         #f)  ;; Generated by a call to (tests:filter-test-names all-test-names test-patts))
	 (required-tests     #f)  ;; Put fully qualified test/testpath names in this list to be done
         (waitors-upon       (make-hash-table)) ;; given a test, return list of tests waiting upon this test.
	 (task-key           (conc (hash-table->alist flags) " " (get-host-name) " " (current-process-id)))
	 ;; (tdbdat             (tasks:open-db))
	 (config-reruns      (let ((x (configf:lookup *configdat* "setup" "reruns")))
			       (if x (string->number x) #f)))
	 (allowed-tests      #f)
	 (runconf            #f))

    ;; check if readonly
    (when readonly-mode
      (debug:print-error 0 *default-log-port* "megatest.db is readonly.  Cannot proceed.")
      (exit 1))

    ;; per user request. If less than 100Meg space on dbdir partition, bail out with error
    ;; this will reduce issues in database corruption
    (common:check-db-dir-and-exit-if-insufficient)

    ;; override the number of reruns from the configs
    ;; this needs to be done at the place where is first runs:run-tests called 
    ;(if (and config-reruns
	  ;   (> run-count config-reruns))
 	  ;(set! run-count config-reruns))

    ;; (if (tasks:need-server run-id)(tasks:start-and-wait-for-server tdbdat run-id 10))

    (let ((sighand (lambda (signum)
		     ;; (signal-mask! signum) ;; to mask or not? seems to cause issues in exiting
		     (bdat-time-to-exit-set! *bdat* #t)
		     (print "Received signal " signum ", cleaning up before exit. Please wait...")
		     (let ((th1 (make-thread (lambda ()
					       ;; (let ((tdbdat (tasks:open-db)))
						 (rmt:tasks-set-state-given-param-key task-key "killed") ;; )
					       (print "Killed by signal " signum ". Exiting")
					       (thread-sleep! 3)
					       (exit))))
			   (th2 (make-thread (lambda ()
					       (thread-sleep! 5)
					       (debug:print 0 *default-log-port* "Done")
					       (exit 4)))))
		       (thread-start! th2)
		       (thread-start! th1)
		       (thread-join! th2)))))
      (set-signal-handler! signal/int sighand)
      (set-signal-handler! signal/term sighand))

    ;; force the starting of a server -- removed BB 17ww28 - no longer needed.
    ;;(debug:print 0 *default-log-port* "waiting on server...")
    ;;(server:start-and-wait *toppath*)
    
    (runs:set-megatest-env-vars run-id inkeys: keys inrunname: runname) ;; these may be needed by the launching process
    (set! runconf (if (common:file-exists? runconfigf)
		      (setup-env-defaults runconfigf run-id *already-seen-runconfig-info* keyvals target)
		      (begin
			(debug:print 0 *default-log-port* "WARNING: You do not have a run config file: " runconfigf)
			#f)))

    (if (not test-patts) ;; first time in - adjust testpatt
	(set! test-patts (common:args-get-testpatt runconf)))
    ;; if test-patts is #f at this point there is something wrong and we need to bail out
    (if (not test-patts)
	(begin
	  (debug:print 0 *default-log-port* "WARNING: there is no test pattern for this run. Exiting now.")
	  (exit 0)))
    
    (if (args:get-arg "-tagexpr")
	(begin
	  (set! allowed-tests (string-join (runs:get-tests-matching-tags (args:get-arg "-tagexpr")) ","))
	  	  (debug:print-info 0 *default-log-port* "filtering initial test list with tagexpr: " (args:get-arg "-tagexpr") " => " allowed-tests)
		  ));; tests will be ANDed with this list

    ;; register this run in monitor.db
    (rmt:tasks-add "run-tests" user target runname test-patts task-key) ;; params)
    (rmt:tasks-set-state-given-param-key task-key "running")

    #;(common:telemetry-log "run-tests"
                          payload:
                          `( (target . ,target)
                             (run-name . ,runname)
                             (test-patts . ,test-patts) ) )

    
    ;; Now generate all the tests lists
    (set! all-tests-registry (tests:get-all))   ;; hash of testname => path-to-test
    (set! all-test-names     (hash-table-keys all-tests-registry))
    ;; filter first for allowed-tests (from -tagexpr) then for test-patts.
    (set! test-names         (tests:filter-test-names
			      (if allowed-tests
				  (tests:filter-test-names all-test-names allowed-tests)
				  all-test-names)
			      test-patts))

    ;; I think seeding required-tests with all test-names makes sense but lack analysis to back that up.

    ;; NEW STRATEGY HERE:
    ;; 1. fill required tests with test-patts
    ;; 2. scan testconfigs and if waitons, itemwait, itempatt calc prior test test-patt
    ;; 3. repeat until all deps propagated
    
    ;; any tests with direct mention in test-patts can be added to required
    ;;(set! required-tests     (lset-intersection equal? (string-split test-patts ",") all-test-names))
    (set! required-tests     (tests:filter-test-names all-test-names test-patts))
    ;;
    ;; (set! required-tests     (lset-intersection equal? test-names all-test-names))
    
    ;; look up all tests matching the comma separated list of globs in
    ;; test-patts (using % as wildcard)

    ;; (set! test-names (delete-duplicates (tests:get-valid-tests *toppath* test-patts)))
    (debug:print-info 0 *default-log-port* "tests search path: " (string-intersperse (tests:get-tests-search-path *configdat*) " "))
    (debug:print-info 0 *default-log-port* "all tests:         " (string-intersperse (sort all-test-names string<) " "))
    (debug:print-info 0 *default-log-port* "test names:        " (string-intersperse (sort test-names string<) " "))
    (debug:print-info 0 *default-log-port* "required tests:    " (string-intersperse (sort required-tests string<) " "))

    ;; on the first pass or call to run-tests set FAILS to NOT_STARTED if
    ;; -keepgoing is specified
    (if (eq? *passnum* 0)
	(begin
	  ;; Is this still necessary? I think not. Unreachable tests are marked as such and 
	  ;; should not cause problems here.
	  ;;
	  ;; have to delete test records where NOT_STARTED since they can cause -keepgoing to 
	  ;; get stuck due to becoming inaccessible from a failed test. I.e. if test B depends 
	  ;; on test A but test B reached the point on being registered as NOT_STARTED and test
	  ;; A failed for some reason then on re-run using -keepgoing the run can never complete.
	  ;;
	  ;; (rmt:general-call 'delete-tests-in-state run-id "NOT_STARTED")
	  
	  ;; Now convert anything in allow-auto-rerun to NOT_STARTED
	  ;;
	  (for-each
	   (lambda (state-status)
	     (let* ((ss-lst (string-split-fields "/" state-status #:infix))
		    (state  (if (> (length ss-lst) 0)(car  ss-lst) #f))
		    (status (if (> (length ss-lst) 1)(cadr ss-lst) #f)))
	       (rmt:set-tests-state-status run-id test-names state status "NOT_STARTED" status)))
	   ;; list of state/status pairs separated by spaces
	   (string-split (or (configf:lookup *configdat* "setup" "allow-auto-rerun") "")))))

    ;; Ensure all tests are registered in the test_meta table
    (runs:update-all-test_meta #f)

    ;; run the run prehook if there are no tests yet run for this run:
    ;;
    (runs:run-pre-hook run-id)
    ;; mark all test launched flag as false in the meta table 
    (rmt:set-var run-id (conc "lunch-complete-" run-id) "no")
    (debug:print-info 1 *default-log-port* "Setting end-of-run to no")
    (let* ((config-reruns      (let ((x (configf:lookup *configdat* "setup" "reruns")))
			       (if x (string->number x) #f)))
	  (config-rerun-cnt (if config-reruns
			config-reruns
			1)))
    (if (eq? config-rerun-cnt run-count)
      (rmt:set-var run-id (conc "end-of-run-" run-id) "no")))

    (rmt:set-run-state-status run-id "new" "n/a")
    ;; now add non-directly referenced dependencies (i.e. waiton)
    ;;======================================================================
    ;; refactoring this block into tests:get-full-data
    ;;
    ;; What happended, this code is now duplicated in tests!?
    ;;
    ;;======================================================================
    
    (if (not (null? test-names)) ;; BEGIN test-names loop
	(let loop ((hed (car test-names))   ;; NOTE: This is the main loop that iterates over the test-names
		   (tal (cdr test-names)))         ;; 'return-procs tells the config reader to prep running system but return a proc
          (debug:print-info 4 *default-log-port* "\n\ntestpatt elaboration loop => hed="hed " tal="tal" test-patts="test-patts" test-names="test-names)
	  (change-directory *toppath*) ;; PLEASE OPTIMIZE ME!!! I think this should be a no-op but there are several places where change-directories could be happening.
	  (set-environment-variable! "MT_TEST_NAME" hed) ;; 
	  (let*-values (((waitons waitors config) (tests:get-waitons hed all-tests-registry))

			;; NOTE: Have the config - can extract [waitons] section
			
                        ((hed-mode)
                         (let ((m (configf:lookup config "requirements" "mode")))
                           (if m (map string->symbol (string-split m)) '(normal))))
                        ((hed-itemized-waiton) ;; are items in hed waiting on items of waiton?
                         (not (null? (lset-intersection eq? hed-mode '(itemmatch itemwait)))))
                        )
	    (debug:print-info 8 *default-log-port* "waitons: " waitons)
	    ;; check for hed in waitons => this would be circular, remove it and issue an
	    ;; error
	    (if (or (member hed waitons)
		    (member hed waitors))
		(begin
		  (debug:print-error 0 *default-log-port* "test " hed " has listed itself as a waiton or waitor, please correct this!")
		  (set! waitons (filter (lambda (x)(not (equal? x hed))) waitons))
		  (set! waitors (filter (lambda (x)(not (equal? x hed))) waitors))))
	    
	    ;; (items   (items:get-items-from-config config)))
	    (if (not (hash-table-ref/default test-records hed #f)) ;; waiton-tconfig below will be #f until that test is visted here at least once
		(hash-table-set! test-records ;; BB: we are doing a manual make-tests:testqueue
				 hed (vector hed     ;; 0 ;; testname
					     config  ;; 1 
					     waitons ;; 2 
					     (configf:lookup config "requirements" "priority")     ;; priority 3
					     (tests:get-items config) ;; 4 ;; expand the [items] and or [itemstable] into explict items
					     #f      ;; itemsdat 5
					     #f      ;; spare - used for item-path
					     waitors ;; 
					     )))
            ;; update waitors-upon here
            (for-each
             (lambda (waiton)
               (let* ((current-waitors-upon (hash-table-ref/default waitors-upon waiton '())))
                 (debug:print-info 8 *default-log-port* " current-waiters-upon["waiton"] is "current-waitors-upon )
                 (when (not (member hed current-waitors-upon))
                   (debug:print-info 8 *default-log-port* " current-waiters-upon["waiton"] << "hed  )
                   (hash-table-set! waitors-upon waiton (cons hed current-waitors-upon)))))
             (if (list? waitons) waitons '()))
            (debug:print-info 8 *default-log-port* " process waitons&waitors of "hed": "(delete-duplicates (append waitons waitors)))
            (for-each 
	     (lambda (waiton)
	       (if (and waiton (not (member waiton test-names)))
		   (let* ((waitors-in-testpatt (runs:testpatts-mention-waitors-upon? test-patts (hash-table-ref/default waitors-upon waiton '())))
                          (waiton-record   (hash-table-ref/default test-records waiton #f))
			  (waiton-tconfig  (if waiton-record (vector-ref waiton-record 1) #f))
			  (waiton-itemized (and waiton-tconfig
						(or (hash-table-ref/default waiton-tconfig "items" #f)
						    (hash-table-ref/default waiton-tconfig "itemstable" #f))))
			  (itemmaps        (tests:get-itemmaps config))  ;; (configf:lookup config "requirements" "itemmap"))
			  (new-test-patts  (tests:extend-test-patts test-patts hed waiton itemmaps hed-itemized-waiton))) 
		     (debug:print-info 0 *default-log-port* "Test " waiton " has " (if waiton-record "a" "no") " waiton-record and" (if waiton-itemized " " " no ") "items")
		     ;; need to account for test-patt here, if I am test "a", selected with a test-patt of "hed/b%"
		     ;; and we are waiting on "waiton" we need to add "waiton/,waiton/b%" to test-patt
		     ;; is this satisfied by merely appending "/" to the waiton name added to the list?
		     ;;
		     ;; This approach causes all of the items in an upstream test to be run 
		     ;; if we have this waiton already processed once we can analzye it for extending
		     ;; tests to be run, since we can't properly process waitons unless they have been
		     ;; initially added we add them again to be processed on second round AND add the hed
		     ;; back in to also be processed on second round
		     (if waiton-tconfig ;; BB: waiter should be in test-patts as well as the waiton have a tconfig.
                         (if waiton-itemized
                             (if waitors-in-testpatt
                                 (begin
                                   (debug:print-info 0 *default-log-port* "New test patts: " new-test-patts ", prev test patts: " test-patts)
                                   (set! test-names (cons waiton test-names)) ;; need to process this one, only add once the waiton tconfig read
                                   (set! required-tests (cons (conc waiton "/") required-tests))
                                   (set! test-patts new-test-patts))
                                 (begin
                                   (debug:print-info 0 *default-log-port* "Waitor(s) not yet on testpatt for " waiton ", setting up to re-process it")
                                   (set! tal (append (cons waiton tal)(list hed)))))
                             (begin
                               (debug:print-info 0 *default-log-port* "Adding non-itemized test " waiton " to required-tests")
                               (set! required-tests (cons waiton required-tests))
                               (set! test-patts new-test-patts)))
			 (begin
			   (debug:print-info 0 *default-log-port* "No testconfig info yet for " waiton ", setting up to re-process it")
			   (set! tal (append (cons waiton tal)(list hed))))) ;; (cons (conc waiton "/") required-tests))
		     ;; NOPE: didn't work. required needs to be plain test names. Try tacking on to test-patts
		     ;;  - doesn't work
		     ;; (set! test-patts (conc test-patts "," waiton "/"))
		     ;; (set! test-names (cons waiton test-names))))) ;; was an append, now a cons
		     )))
	     (delete-duplicates (append waitons waitors)))
	    (let ((remtests (delete-duplicates (append waitons tal))))
              (debug:print-info 8 *default-log-port* " remtests are "remtests)
	      (if (not (null? remtests))
		  (begin
		    ;; (debug:print-info 0 *default-log-port* "Preprocessing continues for " (string-intersperse remtests ", "))
		    (loop (car remtests)(cdr remtests)))))))) ;; END test-names loop

    (if (not (null? required-tests))
	(debug:print-info 1 *default-log-port* "Adding \"" (string-intersperse required-tests " ") "\" to the run queue"))
    ;; NOTE: these are all parent tests, items are not expanded yet.
    (debug:print-info 4 *default-log-port* "test-records=" (hash-table->alist test-records))
    (let ((reglen (configf:lookup *configdat* "setup" "runqueue")))
      (if (> (length (hash-table-keys test-records)) 0)
	  (let* ((keep-going        #t)
		 (run-queue-retries 5)
		 #;(th1        (make-thread (lambda ()
					    (handle-exceptions
						exn
						(begin
						  (print-call-chain)
						  (print " message: " ((condition-property-accessor 'exn 'message) exn)))
					      (runs:run-tests-queue run-id runname test-records keyvals flags test-patts required-tests
								    (any->number reglen) all-tests-registry)))
					  "runs:run-tests-queue"))
		 (th2        (make-thread (lambda ()			 ;; BBQ: why are we visiting ALL runs here?	    
					    ;; (rmt:find-and-mark-incomplete-all-runs))))) CAN'T INTERRUPT IT ...
					    (let ((run-ids (rmt:get-all-run-ids)))
					      (for-each (lambda (run-id)
							  (if keep-going
							      (handle-exceptions
							       exn
							       (debug:print 0 *default-log-port* "error in calling find-and-mark-incomplete for run-id " run-id ", exn=" exn)
							       (runs:find-and-mark-incomplete-and-check-end-of-run run-id #f)))) ;; ovr-deadtime))) ;; could be root of https://hsdes.intel.com/appstore/article/#/220546828/main -- Title: Megatest jobs show DEAD even though they are still running (1.64/27)
							run-ids)))
					  "runs: mark-incompletes")))
	    ;; (thread-start! th1)
	    (thread-start! th2)
	    (thread-join! th2) ;; turn off marking incompletes in parallel. see if it is related to the db locks we are seeing.
	    
	    ;; just do the main stuff in the main thread
	    (runs:run-tests-queue run-id runname test-records keyvals flags test-patts required-tests
								    (any->number reglen) all-tests-registry)
	    (set! keep-going #f)
	    ;; (thread-join! th2)
	    ;; if run-count > 0 call, set -preclean and -rerun STUCK/DEAD
	    (if (> run-count 0) ;; handle reruns
		(begin
		  (if (not (hash-table-ref/default flags "-preclean" #f))
		      (hash-table-set! flags "-preclean" #t))
		  (if (not (hash-table-ref/default flags "-rerun" #f))
		      (hash-table-set! flags "-rerun" "ABORT,STUCK/DEAD,n/a,ZERO_ITEMS"))
		  ;; recursive call to self
		  (runs:run-tests target runname test-patts user flags run-count: (- run-count 1)))
		(runs:end-of-run-check run-id)))
	  (debug:print-info 0 *default-log-port* "No tests to run")))
    (debug:print-info 4 *default-log-port* "All done by here")
    ;; TODO: try putting post hook call here
      
    ;  (debug:print-info 2 *default-log-port* " run-count " run-count)
    ;  (runs:run-post-hook run-id))
    ;  (debug:print-info 2 *default-log-port* "Not calling post hook runcount = " run-count ))   
    (rmt:tasks-set-state-given-param-key task-key "done")
     
    ;; (sqlite3:finalize! tasks-db)
    ))


;; loop logic. These are used in runs:run-tests-queue to make it a bit more readable.
;;
;; If reg not full and have items in tal then loop with (car tal)(cdr tal) reg reruns
;; If reg is full (i.e. length >= n
;;   loop with (car reg) tal (cdr reg) reruns
;; If tal is empty
;;   but have items in reg; loop with (car reg)(cdr reg) '() reruns
;;   If reg is empty => all done

(define (runs:queue-next-hed tal reg n regfull)
  (if regfull
      (if (null? reg) #f (car reg))
      (if (null? tal) ;; tal is used up, pop from reg
	  (if (null? reg) #f (car reg))
	  (car tal))))

(define (runs:queue-next-tal tal reg n regfull)
  (if regfull
      tal
      (if (null? tal) ;; must transfer from reg
	  (if (null? reg) '() (cdr reg))
	  (cdr tal))))

(define (runs:queue-next-reg tal reg n regfull)
  (if regfull
      (if (null? reg) '() (cdr reg)) ;; EXPLORE: reorder (cdr reg) such that looping is more efficient
      (if (null? tal) ;; if tal is null and reg not full then '() as reg contents moved to tal
	  '()
	  reg)))

;; this is the list of parameters to the named loop "loop" near the top of runs:run-tests-queue, look around line 1216
;;
(define (runs:loop-values tal reg reglen regfull reruns)
  (list (runs:queue-next-hed tal reg reglen regfull)      ;; hed
        (runs:queue-next-tal tal reg reglen regfull)      ;; tal
        (runs:queue-next-reg tal reg reglen regfull)      ;; reg
        reruns))                                          ;; reruns

;; objective - iterate thru tests
;;    => want to prioritize tests we haven't seen before
;;    => sometimes need to squeeze things in (added to reg)
;;    => review of a previously seen test is higher priority of never visited test
;; reg - list of previously visited tests
;; tal - list of never visited tests
;;   prefer next hed to be from reg than tal.

(define runs:nothing-left-in-queue-count 0)

;;======================================================================
;; runs:expand-items is called by runs:run-tests-queue
;;======================================================================
;;
;; return value of runs:expand-items is passed back to runs-tests-queue and is fed to named loop with this signature:
;;    (let loop ((hed         (car sorted-test-names))
;;	         (tal         (cdr sorted-test-names))
;;	         (reg         '()) ;; registered, put these at the head of tal 
;;	         (reruns      '()))
(define (runs:expand-items hed tal reg reruns regfull newtal jobgroup max-concurrent-jobs run-id waitons item-path testmode test-record can-run-more items runname tconfig reglen test-registry test-records itemmaps)
  (let* ((loop-list       (list hed tal reg reruns))
	 (prereqs-not-met (let ((res (rmt:get-prereqs-not-met run-id waitons hed item-path mode: testmode itemmaps: itemmaps)))
			    (if (list? res)
				res
				(begin
				  (debug:print 0 *default-log-port*
					       "ERROR: rmt:get-prereqs-not-met returned non-list!\n"
					       "  res=" res " run-id=" run-id " waitons=" waitons " hed=" hed " item-path=" item-path " testmode=" testmode " itemmaps=" itemmaps)
				  '()))))
         (have-itemized (not (null? (lset-intersection eq? testmode '(itemmatch itemwait)))))
	 ;; (prereqs-not-met (mt:lazy-get-prereqs-not-met run-id waitons item-path mode: testmode itemmap: itemmap))
	 (fails           (runs:calc-fails prereqs-not-met))
	 (prereq-fails    (runs:calc-prereq-fail prereqs-not-met))
	 (non-completed   (runs:calc-not-completed prereqs-not-met))
	 (runnables       (runs:calc-runnable prereqs-not-met))
         (unexpanded-prereqs
          (filter (lambda (testname)
                    (let* ((test-rec (hash-table-ref test-records testname))
                           (items       (tests:testqueue-get-items  test-rec)))
                      ;;(BB> "HEY " testname "=>"items)
                      (or (procedure? items)(eq? items 'have-procedure))))
                  waitons))


         )
    (debug:print-info 4 *default-log-port* "START OF INNER COND #2 "
		      "\n can-run-more:    " can-run-more
		      "\n testname:        " hed
		      "\n prereqs-not-met: " (runs:pretty-string prereqs-not-met)
		      "\n non-completed:   " (runs:pretty-string non-completed) 
		      "\n prereq-fails:    " (runs:pretty-string prereq-fails)
		      "\n fails:           " (runs:pretty-string fails)
		      "\n testmode:        " testmode
		      "\n (member 'toplevel testmode): " (member 'toplevel testmode)
		      "\n (null? non-completed):    " (null? non-completed)
		      "\n reruns:          " reruns
		      "\n items:           " items
		      "\n can-run-more:    " can-run-more)

   (cond
     ;; all prereqs met, fire off the test
     ;; or, if it is a 'toplevel test and all prereqs not met are COMPLETED then launch

    ((and (not (member 'toplevel testmode))
	   (member (hash-table-ref/default test-registry (db:test-make-full-name hed item-path) 'n/a)
		   '(DONOTRUN removed CANNOTRUN))) ;; *common:cant-run-states-sym*) ;; '(COMPLETED KILLED WAIVED UNKNOWN INCOMPLETE)) ;; try to catch repeat processing of COMPLETED tests here
      (debug:print-info 4 *default-log-port* "cond branch - "  "ei-1")
      (debug:print-info 1 *default-log-port* "Test " hed " set to \"" (hash-table-ref test-registry (db:test-make-full-name hed item-path)) "\". Removing it from the queue")
      (if (or (not (null? tal))
	      (not (null? reg)))
          (runs:loop-values tal reg reglen regfull reruns)
	  (begin
	    (debug:print-info 0 *default-log-port* "Nothing left in the queue!")
	    ;; If get here twice then we know we've tried to expand all items
	    ;; since there must be a logic issue with the handling of loops in the 
	    ;; items expand phase we will brute force an exit here.
	    (if (> runs:nothing-left-in-queue-count 2)
		(begin
		  (debug:print 0 *default-log-port* "WARNING: this condition is triggered when there were no items to expand and nothing to run. Please check your run for completeness")
		  (exit 0))
		(set! runs:nothing-left-in-queue-count (+ runs:nothing-left-in-queue-count 1)))
	    #f)))

     ;; desired result of below cond branch:
     ;;   we want to expand items in our test of interest (hed) in the following cases:
     ;;    case 1 - mode is itemmatch or itemwait: 
     ;;       - all prereq tests have been expanded
     ;;       - at least one prereq's items have completed
     ;;    case 2 - mode is toplevel   
     ;;       - prereqs are completed.
     ;;       - or no prereqs can complete 
     ;;    case 3 - mode not specified 
     ;;       - prereqs are completed and passed (we could consider removing "and passed" -- it would change behavior from current)          
    ((or (null? prereqs-not-met)
	  (and (member 'toplevel testmode)
	       (null? non-completed)))
      (debug:print-info 4 *default-log-port* "cond branch - "  "ei-2")
      (debug:print-info 4 *default-log-port* "runs:expand-items: (or (null? prereqs-not-met) (and (member 'toplevel testmode)(null? non-completed)))")
      (let ((test-name (tests:testqueue-get-testname test-record)))
	(set-environment-variable! "MT_TEST_NAME" test-name) ;; 
	(set-environment-variable! "MT_RUNNAME"   runname)
	(runs:set-megatest-env-vars run-id inrunname: runname) ;; these may be needed by the launching process
	(let ((items-list (items:get-items-from-config tconfig)))
	  (if (list? items-list)
	      (begin
		(if (null? items-list)
		    (let ((test-id   (rmt:get-test-id run-id test-name ""))
			  (num-items (rmt:test-toplevel-num-items run-id test-name)))
		      (if (and test-id
			       (not (> num-items 0)))
			  (mt:test-set-state-status-by-id run-id test-id "NOT_STARTED" "ZERO_ITEMS" "Failed to run due to failed prerequisites"))))
		(tests:testqueue-set-items! test-record items-list)
		(list hed tal reg reruns))
	      (begin
		(debug:print-error 0 *default-log-port* "The proc from reading the items table did not yield a list - please report this")
		(exit 1))))))

    ((and (null? fails)
	   (null? prereq-fails)
	   (not (null? non-completed)))
      (debug:print-info 4 *default-log-port* "cond branch - "  "ei-3")
      (let* ((allinqueue (map (lambda (x)(if (string? x) x (db:test-get-testname x)))
        		      (append newtal reruns)))
	     ;; prereqstrs is a list of test names as strings that are prereqs for hed
             (prereqstrs (delete-duplicates (map (lambda (x)(if (string? x) x (db:test-get-testname x)))
						 prereqs-not-met)))
	     ;; a prereq that is not found in allinqueue will be put in the notinqueue list
	     ;; 
             ;; (notinqueue (filter (lambda (x)
             ;;    		   (not (member x allinqueue)))
             ;;    		 prereqstrs))
	     (give-up    #f))

	;; We can get here when a prereq has not been run due to *it* having a prereq that failed.
	;; We need to use this to dequeue this item as CANNOTRUN
	;; 
	(if (member 'toplevel testmode) ;; '(toplevel)) ;; NOTE: this probably should be (member 'toplevel testmode)
	    (for-each (lambda (prereq)
			(if (eq? (hash-table-ref/default test-registry prereq 'justfine) 'CANNOTRUN)
			    (set! give-up #t)))
		      prereqstrs))

	(if (and give-up
		 (not (and (null? tal)(null? reg))))
	    (let ((trimmed-tal (mt:discard-blocked-tests run-id hed tal test-records))
		  (trimmed-reg (mt:discard-blocked-tests run-id hed reg test-records)))
	      (debug:print 1 *default-log-port* "WARNING: test " hed " has discarded prerequisites, removing it from the queue")

	      (let ((test-id (rmt:get-test-id run-id hed "")))
		(if test-id (mt:test-set-state-status-by-id run-id test-id "COMPLETED" "PREQ_DISCARDED" "Failed to run due to discarded prerequisites")))
	      
	      (if (and (null? trimmed-tal)
		       (null? trimmed-reg))
		  #f
                  (runs:loop-values trimmed-tal trimmed-reg reglen regfull reruns)
                  ))
	      (list (car newtal)(append (cdr newtal) reg) '() reruns))))

    ((and (null? fails) ;; have not-started tests, but unable to run them.  everything looks completed with no prospect of unsticking something that is stuck.  we should mark hed as moribund and exit or continue if there are more tests to consider
	   (null? prereq-fails)
	   (null? non-completed))
     (debug:print-info 4 *default-log-port* "cond branch - "  "ei-4")
      (if  (runs:can-keep-running? hed 20)
	  (begin
	    (runs:inc-cant-run-tests hed)
	    (debug:print-info 0 *default-log-port* "no fails in prerequisites for " hed " but also none running, keeping " hed " for now. Try count: " (hash-table-ref/default *seen-cant-run-tests* hed 0) ", going to wait 60 sec.") ;; 
	    ;; getting here likely means the system is way overloaded, kill a full minute before continuing
	    ;; (thread-sleep! 60) ;; TODO: gate by normalized server load > 1.0 (maxload config thing) CHECKTHIS!!!
	    ;; No runsdat, can't do this yet
	    ;; (if (runs:dat-load-mgmt-function runsdat)((runs:dat-load-mgmt-function runsdat)))
	    ;;
	    (thread-sleep! 5) ;; TODO: gate by normalized server load > 1.0 (maxload config thing)
	    ;; num-retries code was here
	    ;; we use this opportunity to move contents of reg to tal
	    (list (car newtal)(append (cdr newtal) reg) '() reruns)) ;; an issue with prereqs not yet met?
	  (begin
	    (debug:print-info 1 *default-log-port* "no fails in prerequisites for " hed " but nothing seen running in a while, dropping test " hed " from the run queue")
	    (let ((test-id (rmt:get-test-id run-id hed "")))
	      (if test-id (mt:test-set-state-status-by-id run-id test-id "NOT_STARTED" "TIMED_OUT" "Nothing seen running in a while.")))
            (runs:loop-values tal reg reglen regfull reruns)
            )))

    ((and 
       (or (not (null? fails))
	   (not (null? prereq-fails)))
       (member 'normal testmode))
      (debug:print-info 4 *default-log-port* "cond branch - "  "ei-5")
      (debug:print-info 1 *default-log-port* "test "  hed " (mode=" testmode ") has failed prerequisite(s); "
			(string-intersperse (map (lambda (t)(conc (db:test-get-testname t) ":" (db:test-get-state t)"/"(db:test-get-status t))) fails) ", ")
			", removing it from to-do list")
      (let ((test-id (rmt:get-test-id run-id hed "")))
	(if test-id
	    (if (not (null? prereq-fails))
		(mt:test-set-state-status-by-id-unless-completed run-id test-id "COMPLETED" "PREQ_DISCARDED" "Failed to run due to prior failed prerequisites")
		(mt:test-set-state-status-by-id-unless-completed run-id test-id "COMPLETED" "PREQ_FAIL"      "Failed to run due to failed prerequisites"))))
      ;; (debug:print 4 *default-log-port*"BB> set PREQ_FAIL on "hed)
      ;; (mt:test-set-state-status-by-id run-id test-id "NOT_STARTED" "PREQ_FAIL"      "Failed to run due to failed prerequisites")))) ;; BB: this works, btu equivalent for itemwait mode does not work.
      (if (or (not (null? reg))(not (null? tal)))
	  (begin
	    (hash-table-set! test-registry hed 'CANNOTRUN)
            (runs:loop-values tal reg reglen regfull (cons hed reruns))
            )
	  #f)) ;; #f flags do not loop
      
     ((and (not (null? fails))(member 'toplevel testmode))
      (debug:print-info 4 *default-log-port* "cond branch - "  "ei-6")
      (if (or (not (null? reg))(not (null? tal)))
	   (list (car newtal)(append (cdr newtal) reg) '() reruns)
	  #f)) 
     ((null? runnables)
      (debug:print-info 4 *default-log-port* "cond branch - "  "ei-7")
      #f) ;; if we get here and non-completed is null then it is all over.
     (else
      (debug:print-info 4 *default-log-port* "cond branch - "  "ei-8")
      (debug:print 0 *default-log-port* "WARNING: FAILS or incomplete tests maybe preventing completion of this run. Watch for issues with test " hed ", continuing for now")
      (list (car newtal)(cdr newtal) reg reruns)))))

(define (runs:mixed-list-testname-and-testrec->list-of-strings inlst)
  (if (null? inlst)
      '()
      (map (lambda (t)
	     (cond
	      ((vector? t)
	       (let ((test-name (db:test-get-testname t))
		     (item-path (db:test-get-item-path t))
		     (test-state (db:test-get-state t))
		     (test-status (db:test-get-status t)))
		 (conc test-name (if (equal? item-path "") "" "/") item-path ":" test-state "/" test-status)))
	      ((string? t)
	       t)
	      (else 
	       (conc t))))
	   inlst)))


;;  hed tal reg reruns reglen regfull test-record runname test-name item-path jobgroup max-concurrent-jobs run-id waitons item-path testmode test-patts required-tests test-registry registry-mutex flags keyvals run-info newtal all-tests-registry itemmaps)
(define (runs:process-expanded-tests runsdat testdat)
  ;; unroll the contents of runsdat and testdat (due to ongoing refactoring).
  (debug:print 2 *default-log-port* "runs:process-expanded-tests; testdat:" )
  (debug:print 2 *default-log-port* (with-output-to-string
                                            (lambda () (pp (runs:testdat->alist testdat) ))))
  (let* ((hed                    (runs:testdat-hed testdat))
	 (tal                    (runs:testdat-tal testdat))
	 (reg                    (runs:testdat-reg testdat))
	 (reruns                 (runs:testdat-reruns testdat))
	 (test-name              (runs:testdat-test-name testdat))
	 (item-path              (runs:testdat-item-path testdat))
	 (jobgroup               (runs:testdat-jobgroup testdat))
	 (waitons                (runs:testdat-waitons testdat))
	 (item-path              (runs:testdat-item-path testdat))
	 (testmode               (runs:testdat-testmode testdat))
	 (newtal                 (runs:testdat-newtal testdat))
	 (itemmaps               (runs:testdat-itemmaps testdat))
	 (test-record            (runs:testdat-test-record testdat))
	 (prereqs-not-met        (runs:testdat-prereqs-not-met testdat))

	 (reglen                 (runs:dat-reglen runsdat))
	 (regfull                (runs:dat-regfull runsdat))
	 (runname                (runs:dat-runname runsdat))
	 (max-concurrent-jobs    (runs:dat-max-concurrent-jobs runsdat))
	 (run-id                 (runs:dat-run-id runsdat))
	 (test-patts             (runs:dat-test-patts runsdat))
	 (required-tests         (runs:dat-required-tests runsdat))
	 (test-registry          (runs:dat-test-registry runsdat))
	 (registry-mutex         (runs:dat-registry-mutex runsdat))
	 (flags                  (runs:dat-flags runsdat))
	 (keyvals                (runs:dat-keyvals runsdat))
	 (run-info               (runs:dat-run-info runsdat))
	 (all-tests-registry     (runs:dat-all-tests-registry runsdat))
	 (run-limits-info        (runs:dat-can-run-more-tests runsdat))
	 ;; (runs:can-run-more-tests run-id jobgroup max-concurrent-jobs)) ;; look at the test jobgroup and tot jobs running
	 (have-resources         (car run-limits-info))
	 (num-running            (list-ref run-limits-info 1))
	 (num-running-in-jobgroup(list-ref run-limits-info 2)) 
	 (max-concurrent-jobs    (list-ref run-limits-info 3))
	 (job-group-limit        (list-ref run-limits-info 4))
	 ;; (prereqs-not-met        (rmt:get-prereqs-not-met run-id waitons hed item-path mode: testmode itemmaps: itemmaps))
	 ;; (prereqs-not-met         (mt:lazy-get-prereqs-not-met run-id waitons item-path mode: testmode itemmap: itemmap))
	 (fails                  (if (list? prereqs-not-met) ;; TODO: rename fails to failed-prereqs
				      (runs:calc-fails prereqs-not-met)
				      (begin
					(debug:print-error 0 *default-log-port* "prereqs-not-met is not a list! " prereqs-not-met)
					'())))
	 (non-completed           (filter (lambda (x)             ;; remove hed from not completed list, duh, of course it is not completed!
					    (not (equal? x hed)))
					  (runs:calc-not-completed prereqs-not-met)))
	 (loop-list               (list hed tal reg reruns))
	 ;; configure the load runner
	 (maxload                 (string->number (or (configf:lookup *configdat* "jobtools" "maxload") "3.0")))         ;; use a non-number string to disable
         (maxhomehostload         (string->number (or (configf:lookup *configdat* "jobtools" "maxhomehostload") "2.0"))) ;; use a non-number string to disable
         (waitdelay               (string->number (or (configf:lookup *configdat* "jobtools" "waitdelay") "60"))))
    (debug:print-info 4 *default-log-port* "have-resources: " have-resources " prereqs-not-met: (" 
		      (string-intersperse 
		       (map (lambda (t)
			      (if (vector? t)
				  (conc (db:test-get-state t) "/" (db:test-get-status t))
				  (conc " WARNING: t is not a vector=" t )))
			    prereqs-not-met)
		       ", ") ") fails: " fails
		       "\nregistered? " (hash-table-ref/default test-registry (db:test-make-full-name test-name item-path) #f))
			    
    ;; well, first lets see if cpu load throttling is enabled. If so wait around until the
    ;; average cpu load is under the threshold before continuing
    ;;
    (if (runs:dat-load-mgmt-function runsdat)
	((runs:dat-load-mgmt-function runsdat))
	(runs:dat-load-mgmt-function-set!
	 runsdat
	 (lambda ()
	   ;; jobtools maxload is useful for where the full Megatest run is done on one machine
	   (if (and #;(not (common:on-homehost?))
		    maxload) ;; only gate if maxload is specified, NOTE: maxload is normalized, i.e. load=1 means all cpus fully utilized
	       (common:wait-for-normalized-load maxload "Waiting for load to drop before starting more tests" #f))
	   
	   ;; jobtools maxhomehostload is intended to prevent overloading on the homehost which can cause database corruption issues
	   #;(if maxhomehostload
	       (common:wait-for-homehost-load maxhomehostload
					      (conc "Waiting for homehost load to drop below normalized value of " maxhomehostload))))))
    
 
    
    (if (and (not (null? prereqs-not-met))
	     (runs:lownoise (conc "waiting on tests " prereqs-not-met hed) 60))
	(debug:print-info 2 *default-log-port* "waiting on tests; " (string-intersperse (runs:mixed-list-testname-and-testrec->list-of-strings prereqs-not-met) ", ")))

    ;; Don't know at this time if the test have been launched at some time in the past
    ;; i.e. is this a re-launch?
    (debug:print-info 4 *default-log-port* "run-limits-info = " run-limits-info)
    
    (cond ; cond 894- 1067
     
     ;; Check item path against item-patts, 
     ;;
     ((not (tests:match test-patts (tests:testqueue-get-testname test-record) item-path required: required-tests)) ;; This test/itempath is not to be run
      ;; else the run is stuck, temporarily or permanently
      ;; but should check if it is due to lack of resources vs. prerequisites
      (debug:print-info 1 *default-log-port* "Skipping " (tests:testqueue-get-testname test-record) " " item-path " as it doesn't match " test-patts)
      (if (or (not (null? tal))(not (null? reg)))
	  (runs:loop-values tal reg reglen regfull reruns)
	  #f))
     
     ;; Register tests 
     ;;
     ((not (hash-table-ref/default test-registry (db:test-make-full-name test-name item-path) #f))
      (debug:print-info 4 *default-log-port* "Pre-registering test " test-name "/" item-path " to create placeholder" )
      ;; always do firm registration now in v1.60 and greater ;; (eq? *transport-type* 'fs) ;; no point in parallel registration if use fs
      (let register-loop ((numtries 15))
	(rmt:register-test run-id test-name item-path)
	(if (rmt:get-test-id run-id test-name item-path)
	    (hash-table-set! test-registry (db:test-make-full-name test-name item-path) 'done)
	    (if (> numtries 0)
		(begin
		  (thread-sleep! 0.5)
		  (register-loop (- numtries 1)))
		(debug:print-error 0 *default-log-port* "failed to register test " (db:test-make-full-name test-name item-path)))))
      (if (not (eq? (hash-table-ref/default test-registry (db:test-make-full-name test-name "") #f) 'done))
	  (begin
	    (rmt:register-test run-id test-name "")
	    (if (rmt:get-test-id run-id test-name "")
		(hash-table-set! test-registry (db:test-make-full-name test-name "") 'done))))
      (runs:shrink-can-run-more-tests-count runsdat)   ;; DELAY TWEAKER (still needed?)
      (if (and (null? tal)(null? reg))
	  (list hed tal (append reg (list hed)) reruns)
	  (list (runs:queue-next-hed tal reg reglen regfull) ;; cannot replace with a call to runs:loop-values as the logic is different for reg
		(runs:queue-next-tal tal reg reglen regfull)
		;; NB// Here we are building reg as we register tests
		;; if regfull we must pop the front item off reg
		(if regfull
		    (append (cdr reg) (list hed))
		    (append reg (list hed)))
		reruns)))
     
     ;; At this point hed test registration must be completed.
     ;;
     ((eq? (hash-table-ref/default test-registry (db:test-make-full-name test-name item-path) #f)
	   'start)
      (debug:print-info 0 *default-log-port* "Waiting on test registration(s): "
			(string-intersperse 
			 (filter (lambda (x)
				   (eq? (hash-table-ref/default test-registry x #f) 'start))
				 (hash-table-keys test-registry))
			 ", "))
      (thread-sleep! 0.051)
      (list hed tal reg reruns))
     
     ;; If no resources are available just kill time and loop again
     ;;
     ((not have-resources) ;; simply try again after waiting a second
      (if (runs:lownoise "no resources" 60)
	  (debug:print-info 1 *default-log-port* "no resources to run new tests, waiting ..."))
      ;; Have gone back and forth on this but db starvation is an issue.
      ;; wait one second before looking again to run jobs.
      (thread-sleep! 0.253)
      ;; could have done hed tal here but doing car/cdr of newtal to rotate tests
      (list (car newtal)(cdr newtal) reg reruns))
     
     ;; This is the final stage, everything is in place so launch the test
     ;;
     ((and have-resources
	   (or (null? prereqs-not-met)
	       (and (member 'toplevel testmode) ;;  'toplevel)
		    (null? non-completed)
		    (not (member 'exclusive testmode)))))
      ;; (hash-table-delete! *max-tries-hash* (db:test-make-full-name test-name item-path))
      ;; we are going to reset all the counters for test retries by setting a new hash table
      ;; this means they will increment only when nothing can be run
      (set! *max-tries-hash* (make-hash-table))
      
      (run:test run-id run-info keyvals runname test-record flags #f test-registry all-tests-registry runsdat testdat)
      (runs:incremental-print-results run-id)
      (hash-table-set! test-registry (db:test-make-full-name test-name item-path) 'running)
      (runs:shrink-can-run-more-tests-count runsdat)  ;; DELAY TWEAKER (still needed?)
      ;; (thread-sleep! *global-delta*)
      (if (or (not (null? tal))(not (null? reg)))
	  (runs:loop-values tal reg reglen regfull reruns) ;; hed should be dropped at this time
	  #f))
     
     ;; must be we have unmet prerequisites
     ;;
     (else
      (debug:print 4 *default-log-port* "FAILS: " fails)
      ;; If one or more of the prereqs-not-met are FAIL then we can issue
      ;; a message and drop hed from the items to be processed.
      ;; (runs:mixed-list-testname-and-testrec->list-of-strings prereqs-not-met)
      (if (and (not (null? prereqs-not-met))
	       (runs:lownoise (conc "waiting on tests " prereqs-not-met hed) 60))
	  (debug:print-info 1 *default-log-port* "waiting on tests; " (string-intersperse 
						    (runs:mixed-list-testname-and-testrec->list-of-strings 
						     prereqs-not-met) ", ")))
      (if (or (null? fails)
	      (member 'toplevel testmode))
	  (begin
	    ;; couldn't run, take a breather
	    (if  (runs:lownoise "Waiting for more work to do..." 60)
		 (debug:print-info 0 *default-log-port* "Waiting for more work to do..."))
	    
	    ;; (if (runs:dat-load-mgmt-function runsdat)((runs:dat-load-mgmt-function runsdat)))
	    (thread-sleep! 5)
	    (list (car newtal)(cdr newtal) reg reruns))
	  ;; the waiton is FAIL so no point in trying to run hed ever again
	  (begin
            (let ((my-test-id (rmt:get-test-id run-id test-name item-path)))
              (mt:test-set-state-status-by-id-unless-completed run-id my-test-id "COMPLETED" "PREQ_FAIL" "Failed to run due to failed prerequisites2"))            
            (if (or (not (null? reg))(not (null? tal)))
                (if (vector? hed)
		  (begin
		    (debug:print 1 *default-log-port* "WARNING: Dropping test " test-name "/" item-path
				 " from the launch list as it has prerequistes that are FAIL")
		    (let ((test-id (rmt:get-test-id run-id hed "")))
		      (if test-id (mt:test-set-state-status-by-id-unless-completed run-id test-id "COMPLETED" "PREQ_FAIL" "Failed to run due to failed prerequisites")))
		    (runs:shrink-can-run-more-tests-count runsdat) ;; DELAY TWEAKER (still needed?)
		    ;; (thread-sleep! *global-delta*)
		    ;; This next is for the items

                    (if (not (null? fails))
                        ;;(mt:test-set-state-status-by-testname run-id test-name item-path "NOT_STARTED" "PREQ_FAIL" #f)
                        (rmt:set-state-status-and-roll-up-items run-id test-name item-path "NOT_STARTED" "PREQ_FAIL" #f) 
                        ;;(mt:test-set-state-status-by-testname run-id test-name item-path "NOT_STARTED" "BLOCKED" #f)
                        (rmt:set-state-status-and-roll-up-items run-id test-name item-path "NOT_STARTED" "BLOCKED" #f) )
		    (hash-table-set! test-registry (db:test-make-full-name test-name item-path) 'removed)
		    (runs:loop-values tal reg reglen regfull reruns))
		  (let ((nth-try (hash-table-ref/default test-registry hed 0))) ;; hed not a vector...
                    (debug:print 2 *default-log-port* "nth-try("hed")="nth-try)
		    (cond
		     ((member "RUNNING" (map db:test-get-state prereqs-not-met))
		      (if (runs:lownoise (conc "possible RUNNING prerequistes " hed) 60)
			  (debug:print 0 *default-log-port* "WARNING: test " hed " has possible RUNNING prerequisites, don't give up on it yet."))
		      (thread-sleep! 0.1)
		      (runs:loop-values tal reg reglen regfull reruns))
		     ((or (not nth-try) ;; BB: condition on subsequent tries, condition below fires on first try 
			  (and (number? nth-try)
			       (< nth-try 2)))
		      (hash-table-set! test-registry hed (if (number? nth-try)
							     (+ nth-try 1)
							     0))
		      (if (runs:lownoise (conc "not removing test " hed) 60)
			  (debug:print 1 *default-log-port* "WARNING: not removing test " hed " from queue although it may not be runnable due to FAILED prerequisites"))
		      ;; may not have processed correctly. Could be a race condition in your test implementation? Dropping test " hed) ;;  " as it has prerequistes that are FAIL. (NOTE: hed is not a vector)")
		      (runs:shrink-can-run-more-tests-count runsdat) ;; DELAY TWEAKER (still needed?)
		      (runs:loop-values newtal reg reglen regfull reruns))
		     ((symbol? nth-try) ;; BB: 'done matches here in one case where prereq itemwait failed.  This is first "try"
		      (if (eq? nth-try 'removed) ;; removed is removed - drop it NOW
			  (if (null? tal)
			      #f ;; yes, really
			      (list (car tal)(cdr tal) reg reruns))
			  (begin
			    (if (runs:lownoise (conc "FAILED prerequisites or other issue" hed) 60)
				(debug:print 0 *default-log-port* "WARNING: test " hed " has FAILED prerequisites or other issue. Internal state >" nth-try "< will be overridden and we'll retry."))
			    ;; was: (mt:test-set-state-status-by-testname run-id test-name item-path "NOT_STARTED" "KEEP_TRYING" #f)
                            (mt:test-set-state-status-by-testname-unless-completed run-id test-name item-path "COMPLETED" "PREQ_FAIL" #f)
			    (hash-table-set! test-registry hed 'removed) ;; was 0
                            (if (not (and (null? reg) (null? tal)))
                                (runs:loop-values tal reg reglen regfull reruns)
                                #f))))
		     (else
		      (if (runs:lownoise (conc "FAILED prerequitests and we tried" hed) 60)
			  (debug:print 0 *default-log-port* "WARNING: test " hed " has FAILED prerequitests and we've tried at least 10 times to run it. Giving up now."))
		      ;; (debug:print 0 *default-log-port* "         prereqs: " prereqs-not-met)
		      (hash-table-set! test-registry hed 'removed)
		      (mt:test-set-state-status-by-testname run-id test-name item-path "NOT_STARTED" "TEN_STRIKES" #f)
		      ;; I'm unclear on if this roll up is needed - it may be the root cause of the "all set to FAIL" bug.
		      (rmt:set-state-status-and-roll-up-items run-id test-name item-path #f "FAIL" #f) ;; treat as FAIL
		      (list (if (null? tal)(car newtal)(car tal))
			    tal
			    reg
			    reruns)))))
                ;; ELSE: can't drop this - maybe running? Just keep trying

                ;;(if (not (or (not (null? reg))(not (null? tal)))) ;; old experiment
                (let ((runable-tests (runs:runable-tests prereqs-not-met))) ;; SUSPICIOUS: Should look at more than just prereqs-not-met?
                  (if (null? runable-tests)
                      #f   ;; I think we are truly done here
                      (runs:loop-values newtal reg reglen regfull reruns)))
                ;;) ;;from old experiment
                ) ;; end if (or (not (null? reg))(not (null? tal)))
            ))))))

;; scan a list of tests looking to see if any are potentially runnable
;;
(define (runs:runable-tests tests)
  (filter (lambda (t)
	    (if (not (vector? t))
		t
		(let ((state  (db:test-get-state t))
		      (status (db:test-get-status t)))
		  (case (string->symbol state)
		    ((COMPLETED INCOMPLETE) #f)
		    ((NOT_STARTED)
		     (if (member status '("TEN_STRIKES" "BLOCKED" "PREQ_FAIL" "ZERO_ITEMS" "PREQ_DISCARDED" "TIMED_OUT" ))
			 #f
			 t))
		    ((DELETED) #f)
		    (else t)))))
	  tests))

(define (runs:incremental-print-results run-id)
  (let* ((curr-sec    (current-seconds))
	 (runs-data   (bdat-runs-data *bdat*))
	 (last-update (runs:gendat-inc-results-last-update runs-data))
	 (runs-data   (bdat-runs-data *bdat*)))
    (if (> (- curr-sec last-update) 5) ;; at least five seconds since last update
	(let* ((run-dat  (or (runs:gendat-run-info runs-data)(rmt:get-run-info run-id)))
	       (runname  (or (runs:gendat-runname runs-data)
			     (db:get-value-by-header (db:get-rows run-dat)
						     (db:get-header run-dat) "runname")))
	       (target   (or (runs:gendat-target runs-data)(rmt:get-target run-id)))
	       (testsdat (let ((res (rmt:get-tests-for-run
				     run-id "%" '() '() ;; run-id testpatt states statuses
				     #f #f ;; offset limit
				     #f ;; not-in
				     #f ;; sort-by
				     #f ;; sort-order
				     #f ;; get full data (not 'shortlist)
				     last-update
				     'dashboard)))
			   (if (list? res)
			       res
			       (begin
				 (debug:print-error
				  0 *default-log-port*
				  "FAILED TO GET DATA using rmt:get-tests-for-run. Notify developers if you see this. result: " res)
				 '())))))
	  (runs:gendat-inc-results-last-update-set! runs-data (- curr-sec 1))
	  (if (not (runs:gendat-run-info runs-data))
	      (runs:gendat-run-info-set! runs-data run-dat))
	  (if (not (runs:gendat-runname  runs-data))
	      (runs:gendat-runname-set! runs-data runname))
	  (if (not (runs:gendat-target runs-data))
	      (runs:gendat-target-set! runs-data target))
	  (for-each
	   (lambda (testdat)
	     (let* ((test-id    (db:test-get-id           testdat))
		    (prevdat    (hash-table-ref/default   (runs:gendat-inc-results runs-data)
							  (conc run-id "," test-id) #f))
		    (test-name  (db:test-get-testname     testdat))
		    (item-path  (db:test-get-item-path    testdat))
		    (state      (db:test-get-state        testdat))
		    (status     (db:test-get-status       testdat))
		    (event-time (db:test-get-event_time   testdat))
		    (duration   (db:test-get-run_duration testdat)))
	       (if (and (not (member state '("DELETED" "REMOTEHOSTSTART" "RUNNING" "LAUNCHED""NOT_STARTED")))
			(not (and prevdat
				  (equal? state  (db:test-get-state  prevdat))
				  (equal? status (db:test-get-status prevdat)))))
		   (let ((fmt   (runs:gendat-inc-results-fmt runs-data))
			 (dtime (seconds->year-work-week/day-time event-time))) 
		     (if (runs:lownoise "inc-print" 600)
			 (begin
			   (print "fmt=" fmt)
			   (format #t fmt "State" "Status" "Start Time" "Duration" "Test path")))
		     ;; (debug:print 0 *default-log-port* "fmt: " fmt " state: " state " status: " status " test-name: " test-name " item-path: " item-path " dtime: " dtime)
		     ;; (debug:print 0 #f "event-time: " event-time " duration: " duration)
		     (format #t fmt
			     state
			     status
			     dtime
			     (seconds->hr-min-sec duration)
			     (conc "lt/" target "/" runname "/" test-name (if (string-null? item-path) "" (conc "/" item-path))))
		     (hash-table-set! (runs:gendat-inc-results runs-data) (conc run-id "," test-id) testdat)))))
	   testsdat)))

    ;; I don't think this should be here? -- Matt
    #;(runs:gendat-inc-results-last-update-set! runs-data (- curr-sec 10))

    ))

;; every time though the loop increment the test/itempatt val.
;; when the min is > max-allowed and none running then force exit
;;

(define (runs:pretty-long-list lst)
   (if (> (length lst) 8)(append (take lst 3)(list "...")) lst))

;;======================================================================
;; runs:run-tests-queue is called by runs:run-tests
;;======================================================================
;;
;; test-records is a hash table testname:item_path => vector < testname testconfig waitons priority items-info ... >
(define (runs:run-tests-queue run-id runname test-records keyvals flags test-patts required-tests reglen-in all-tests-registry)
  ;; At this point the list of parent tests is expanded 
  ;; NB// Should expand items here and then insert into the run queue.
  (debug:print 5 *default-log-port* "test-records: " test-records ", flags: " (hash-table->alist flags))

  ;; Do mark-and-find clean up of db before starting runing of quue
  ;;
  ;; (rmt:find-and-mark-incomplete)

  (let* ((run-info              (rmt:get-run-info run-id))
         (tests-info            (mt:get-tests-for-run run-id #f '() '())) ;;  qryvals: "id,testname,item_path"))
         (sorted-test-names     (tests:sort-by-priority-and-waiton test-records))
         (test-registry         (make-hash-table))
         (registry-mutex        (make-mutex))
         (num-retries           0)
         (max-retries           (configf:lookup *configdat* "setup" "maxretries"))
         (max-concurrent-jobs   (configf:lookup-number *configdat* "setup" "max_concurrent_jobs" default: 50))
         (reglen                (if (number? reglen-in) reglen-in 1))
         (last-time-incomplete  (- (current-seconds) 900)) ;; force at least one clean up cycle
         (last-time-some-running (current-seconds))
         ;; (tdbdat                (tasks:open-db))
         (runsdat (make-runs:dat
                   ;; hed: hed
                   ;; tal: tal
                   ;; reg: reg
                   ;; reruns: reruns
                   reglen: reglen
                   regfull: #f ;; regfull
                   ;; test-record: test-record
                   runname: runname
                   ;; test-name: test-name
                   ;; item-path: item-path
                   ;; jobgroup: jobgroup
                   max-concurrent-jobs: max-concurrent-jobs
                   run-id: run-id
                   ;; waitons: waitons
                   ;; testmode: testmode
                   test-patts: test-patts
                   required-tests: required-tests
                   test-registry: test-registry
                   registry-mutex: registry-mutex
                   flags: flags
                   keyvals: keyvals
                   run-info: run-info
                   ;; newtal: newtal
                   all-tests-registry: all-tests-registry
                   ;; itemmaps: itemmaps
                   ;; prereqs-not-met: (rmt:get-prereqs-not-met run-id waitons hed item-path mode: testmode itemmaps: itemmaps)
                   ;; can-run-more-tests: (runs:can-run-more-tests run-id jobgroup max-concurrent-jobs) ;; look at the test jobgroup and tot jobs running
                   )))

    ;; Initialize the test-registery hash with tests that already have a record
    ;; convert state to symbol and use that as the hash value
    (for-each (lambda (trec)
		(let ((id (db:test-get-id        trec))
		      (tn (db:test-get-testname  trec))
		      (ip (db:test-get-item-path trec))
		      (st (db:test-get-state     trec)))
		  (if (not (equal? st "DELETED"))
		      (hash-table-set! test-registry (db:test-make-full-name tn ip) (string->symbol st)))))
	      tests-info)
    (set! max-retries (if (and max-retries (string->number max-retries))(string->number max-retries) 100))

    (let loop ((hed         (car sorted-test-names))
               (tal         (cdr sorted-test-names))
	       (reg         '()) ;; registered, put these at the head of tal 
	       (reruns      '()))

      (runs:incremental-print-results run-id)

      (if (not (null? reruns))(debug:print-info 4 *default-log-port* "reruns=" reruns))

      ;; Here we mark any old defunct tests as incomplete. Do this every fifteen minutes
      ;; moving this to a parallel thread and just run it once.
      ;;
      (if (> (current-seconds)(+ last-time-incomplete 900))
          (begin
            (set! last-time-incomplete (current-seconds))
            ;; (rmt:find-and-mark-incomplete-all-runs)
	    ))

      ;; (print "Top of loop, hed=" hed ", tal=" tal " ,reruns=" reruns)
      (let* ((test-record (hash-table-ref test-records hed))
	     (test-name   (tests:testqueue-get-testname test-record))
	     (tconfig     (tests:testqueue-get-testconfig test-record))
	     (jobgroup    (configf:lookup tconfig "test_meta" "jobgroup"))
	     (testmode    (let ((m (configf:lookup tconfig "requirements" "mode")))
			    (if m (map string->symbol (string-split m)) '(normal))))
	     (itemmaps    (tests:get-itemmaps tconfig)) ;;  (configf:lookup tconfig "requirements" "itemmap"))
	     (priority    (tests:testqueue-get-priority   test-record))
	     (itemdat     (tests:testqueue-get-itemdat    test-record)) ;; itemdat can be a string, list or #f
	     (items       (tests:testqueue-get-items      test-record))
	     (item-path   (item-list->path itemdat))
	     (tfullname   (db:test-make-full-name test-name item-path))
	      ;; these are hard coded item-item waits test/item-path => test/item-path2 ...
	     (extra-waits (let* ((section (configf:get-section (tests:testqueue-get-testconfig test-record) "waitons"))
				 (myextra (alist-ref tfullname section equal?)))
			    (if myextra
				(let ((extras (string-split (car myextra))))
				  (if (runs:lownoise (conc tfullname "extra-waitons" tfullname) 60)
				      (debug:print-info 0 *default-log-port* "HAVE EXTRA WAITONS for test " tfullname ": " myextra))
				  (for-each
				   (lambda (extra)
				     ;; (debug:print 0 *default-log-port* "FYI: extra = " extra " reruns = " reruns)
				     (let ((basetestname (car (string-split extra "/"))))
				       #;(if (not (member extra tal))
					   (set! reruns (append tal (list extra))))
				       (if (not (member basetestname tal))
					   (set! reruns (append tal (list basetestname))))
				       ))
				   extras)
				  extras)
				'())))
	     (waitons     (delete-duplicates (append (tests:testqueue-get-waitons test-record) extra-waits) equal?))
	     (newtal      (append tal (list hed)))
	     (regfull     (>= (length reg) reglen))
	     (num-running (rmt:get-count-tests-running-for-run-id run-id))
	     (testdat     (make-runs:testdat
			   hed: hed
			   tal: tal
			   reg: reg
			   reruns: reruns
			   test-record: test-record
			   test-name:   test-name
			   item-path:   item-path
			   jobgroup:    jobgroup
			   waitons:     waitons
			   testmode:    testmode
			   newtal:      newtal
			   itemmaps:    itemmaps
			   ;; prereqs-not-met: prereqs-not-met
			   )))
	(runs:dat-regfull-set! runsdat regfull)
    
	(if (> num-running 0)
            (set! last-time-some-running (current-seconds)))

        (if (> (current-seconds)(+ last-time-some-running (or (configf:lookup *configdat* "setup" "give-up-waiting") 36000)))
            (hash-table-set! *max-tries-hash* tfullname (+ (hash-table-ref/default *max-tries-hash* tfullname 0) 1)))
	;; (debug:print 0 *default-log-port* "max-tries-hash: " (hash-table->alist *max-tries-hash*))

	;; Ensure all top level tests get registered. This way they show up as "NOT_STARTED" on the dashboard
	;; and it is clear they *should* have run but did not.
	(if (not (hash-table-ref/default test-registry (db:test-make-full-name test-name "") #f))
	    (begin
	      (rmt:register-test run-id test-name "")
	      (hash-table-set! test-registry (db:test-make-full-name test-name "") 'done)))
	
	;; Fast skip of tests that are already "COMPLETED" - NO! Cannot do that as the items may not have been expanded yet :(
	;;
	(if (member (hash-table-ref/default test-registry tfullname #f) 
		    '(DONOTRUN removed)) ;; *common:cant-run-states-sym*) ;; '(COMPLETED KILLED WAIVED UNKNOWN INCOMPLETE))
	    (begin
	      (if (runs:lownoise (conc "been marked do not run " tfullname) 60)
		  (debug:print-info 0 *default-log-port* "Skipping test " tfullname " as it has been marked do not run due to being completed or not runnable"))
	      (if (or (not (null? tal))(not (null? reg)))
		  (loop (runs:queue-next-hed tal reg reglen regfull)
			(runs:queue-next-tal tal reg reglen regfull)
			(runs:queue-next-reg tal reg reglen regfull)
			reruns))))
        ;; (loop (car tal)(cdr tal) reg reruns))))

	(runs:incremental-print-results run-id)
	(debug:print 4 *default-log-port* "TOP OF LOOP => "
		     "test-name: " test-name
		     "\n  hed:         " hed
		     "\n  tal:         " (runs:pretty-long-list tal)
		     "\n  reg:         " reg
                     "\n  test-record  " test-record
                     "\n  itemdat:     " itemdat
		     "\n  items:       " items
		     "\n  item-path:   " item-path
		     "\n  waitons:     " waitons
		     "\n  num-retries: " num-retries
                     "\n  reruns:      " reruns
		     "\n  regfull:     " regfull
		     "\n  reglen:      " reglen
		     "\n  length reg:  " (length reg)
                     )

	;; (runs:parallel-runners-mgmt runsdat)

	;; check for hed in waitons => this would be circular, remove it and issue an
	;; error
	(if (member test-name waitons)
	    (begin
	      (debug:print-error 0 *default-log-port* "test " test-name " has listed itself as a waiton, please correct this!")
	      (set! waiton (filter (lambda (x)(not (equal? x hed))) waitons))))

	(cond 
	 
	 ;; We want to catch tests that have waitons that are NOT in the queue and discard them IFF 
	 ;; they have been through the wringer 10 or more times
	 ((and (list? waitons)
	       (not (null? waitons))
	       (> (hash-table-ref/default *max-tries-hash* tfullname 0) 10)
	       (not (null? (filter
			    number?
			    (map (lambda (waiton)
				   (if (and (not (member waiton tal))            ;; this waiton is not in the list to be tried to run
					    (not (member waiton reruns)))
				       1
				       #f))
				 waitons))))) ;; could do this more elegantly with a marker....
          (debug:print-info 4 *default-log-port* "cond branch - "  "rtq-1")
	  (debug:print 0 *default-log-port* "WARNING: Marking test " tfullname " as not runnable. It is waiting on tests that cannot be run. Giving up now.")
	  (hash-table-set! test-registry tfullname 'removed))

	 ;; items is #f then the test is ok to be handed off to launch (but not before)
	 ;; 
	 ((not items)
          (debug:print-info 4 *default-log-port* "cond branch - "  "rtq-2")
	  (debug:print-info 4 *default-log-port* "OUTER COND: (not items)")
	  (if (and (not (tests:match test-patts (tests:testqueue-get-testname test-record) item-path required: required-tests))
		   (not (null? tal)))
	      (loop (car tal)(cdr tal) reg reruns))

	  ;; gonna try a strategy change here.
	  ;;
	  ;; check if can run more tests. if yes, continue, if no, rest until can run more
	  ;; look at the test jobgroup and tot jobs running
	  ;;
	  ;; NOTE: This does NOT actually gate here, only captures the proc to be called later
	  ;; 
	  (if (not (runs:dat-wait-for-jobs-function runsdat))
	      (runs:dat-wait-for-jobs-function-set!
	       runsdat 
	       (lambda (testdat-in)
		 (let* ((jobgroup              (runs:testdat-jobgroup testdat-in))
			(can-run-more-tests    (runs:dat-can-run-more-tests runsdat))
			(last-jobs-check-time  (runs:dat-last-jobs-check-time runsdat))
			(should-check-jobs     (match can-run-more-tests
						 ((can-run-more-flag num-running nr-in-jobgroup max-concurrent-jobs . params)
						  (if (< (- max-concurrent-jobs num-running) 25)
						      (begin
							(debug:print-info 0 *default-log-port*
									  "less than 20 jobs headroom, ("max-concurrent-jobs
									  "-"num-running")>20. Forcing prelaunch check.")
							#t)
						      #f))
						 (else #f)))) ;; no record yet
		   (if should-check-jobs
		       (let loop-can-run-more
			   ((res      (runs:can-run-more-tests runsdat run-id jobgroup max-concurrent-jobs))
			    (remtries 1440)) ;; we can wait for up to two hours for jobs to get done
			 (match res
			   ((run-more num-running . rem)
			    (if (or run-more
				    (< remtries 1))
				(begin
				  (if (runs:lownoise "num-running" 30)
				      (debug:print-info 0 *default-log-port* "Have "num-running" tests of max " max-concurrent-jobs))
				  (runs:dat-can-run-more-tests-set! runsdat res)) ;; capture the result and then drop through
				(begin
				  (if (runs:lownoise "num-running" 10)
				      (debug:print-info 0 *default-log-port* "Can't run more tests, have "num-running" tests of "
							max-concurrent-jobs " allowed."))
				  (thread-sleep! 5) ;; if we've hit max concurrent jobs take a breather, nb// make this configurable
				  
				  ;; wait for load here
				  (if (runs:dat-load-mgmt-function runsdat)((runs:dat-load-mgmt-function runsdat)))
				  (loop-can-run-more (runs:can-run-more-tests runsdat run-id jobgroup max-concurrent-jobs)
						     (- remtries 1)))))))
		       )))))

	  ;; I'm not clear on why prereqs are gathered here TODO: verfiy this is needed
	  (runs:testdat-prereqs-not-met-set! testdat (rmt:get-prereqs-not-met run-id waitons hed item-path mode: testmode itemmaps: itemmaps))

	  ;; I'm not clear on why we'd capture running job counts here TODO: verify this is needed
	  (runs:dat-can-run-more-tests-set! runsdat (runs:can-run-more-tests runsdat run-id jobgroup max-concurrent-jobs))

	  (let ((loop-list (runs:process-expanded-tests runsdat testdat))) ;; in process-expanded-tests ultimately run:test -> launch-test -> test actually running
            (if loop-list (apply loop loop-list))))

	 ;; items processed into a list but not came in as a list been processed
	 ;;
	 ((and (list? items)     ;; thus we know our items are already calculated
	       (not   itemdat))  ;; and not yet expanded into the list of things to be done
          (debug:print-info 4 *default-log-port* "cond branch - "  "rtq-3")
	  (debug:print-info 4 *default-log-port* "OUTER COND: (and (list? items)(not itemdat))")
	  ;; Must determine if the items list is valid. Discard the test if it is not.
	  (if (and (list? items)
		   (> (length items) 0)
		   (and (list? (car items))
			(> (length (car items)) 0))
		   (debug:debug-mode 1))
	      (debug:print 2 *default-log-port* (map (lambda (row)
                                                       (conc (string-intersperse
                                                              (map (lambda (varval)
                                                                     (string-intersperse varval "="))
                                                                   row)
                                                              " ")
                                                             "\n"))
                                                     items)))

          (let* ((items-in-testpatt
                  (filter
                   (lambda (my-itemdat)
                     (tests:match test-patts hed (item-list->path my-itemdat) ))
                   ;; was: (tests:match test-patts hed (item-list->path my-itemdat) required: required-tests))
                   items) ))
            (if (null? items-in-testpatt)
                (debug:print-error 0 *default-log-port* "Test " (tests:testqueue-get-testname test-record) " is itemized but has no items matching the test pattern")
                
                (for-each (lambda (my-itemdat)
                            (let* ((new-test-record (let ((newrec (make-tests:testqueue)))
                                                      (vector-copy! test-record newrec)
                                                      newrec))
                                   (my-item-path (item-list->path my-itemdat))

                                   (newtestname (db:test-make-full-name hed my-item-path)))    ;; test names are unique on testname/item-path
                              (tests:testqueue-set-items!     new-test-record #f)
                              (tests:testqueue-set-itemdat!   new-test-record my-itemdat)
                              (tests:testqueue-set-item_path! new-test-record my-item-path)
                              (hash-table-set! test-records newtestname new-test-record)
                              (set! tal (append tal (list newtestname)))))  ;; since these are itemized create new test names testname/itempath
                          items-in-testpatt)))
          
          

	  ;; At this point we have possibly added items to tal but all must be handed off to 
	  ;; INNER COND logic. I think loop without rotating the queue 
	  ;; (loop hed tal reg reruns))
	  ;; (let ((newtal (append tal (list hed))))  ;; We should discard hed as it has been expanded into it's items? Yes, but only if this *is* an itemized test
	  ;; (loop (car newtal)(cdr newtal) reg reruns)
	  (if (null? tal)
	      #f
	      (loop (car tal)(cdr tal) reg reruns)))
         
	 ;; if items is a proc then need to run items:get-items-from-config, get the list and loop 
	 ;;    - but only do that if resources exist to kick off the job
	 ;; EXPAND ITEMS
	 ((or (procedure? items)(eq? items 'have-procedure))
          (debug:print-info 4 *default-log-port* "cond branch - "  "rtq-4")
	  (let ((can-run-more    #f)) ;; (runs:can-run-more-tests runsdat run-id jobgroup max-concurrent-jobs)))
	    (if (not can-run-more) #;(and (list? can-run-more)
		(car can-run-more))
		(let ((loop-list (runs:expand-items hed tal reg reruns regfull newtal jobgroup max-concurrent-jobs run-id waitons item-path testmode test-record can-run-more items runname tconfig reglen test-registry test-records itemmaps))) ;; itemized test expanded here
		  (if loop-list
		      (apply loop loop-list)
                      (debug:print-info 4 *default-log-port* " -- Can't expand hed="hed)
                      )
                  )
		;; if can't run more just loop with next possible test
		(loop (car newtal)(cdr newtal) reg reruns))))
         
	 ;; this case should not happen, added to help catch any bugs
	 ((and (list? items) itemdat)
          (debug:print-info 4 *default-log-port* "cond branch - "  "rtq-5")
	  (debug:print-error 0 *default-log-port* "Should not have a list of items in a test and the itemspath set - please report this")
	  (exit 1))
	 ((not (null? reruns))
          (debug:print-info 4 *default-log-port* "cond branch - "  "rtq-6")
	  (let* ((newlst (tests:filter-non-runnable run-id tal test-records)) ;; i.e. not FAIL, WAIVED, INCOMPLETE, PASS, KILLED,
		 (junked (lset-difference equal? tal newlst)))
	    (debug:print-info 4 *default-log-port* "full drop through, if reruns is less than 100 we will force retry them, reruns=" reruns ", tal=" tal)
	    (if (< num-retries max-retries)
		(set! newlst (append reruns newlst)))
	    (set! num-retries (+ num-retries 1))
	    ;; (thread-sleep! (+ 1 *global-delta*))
	    (if (not (null? newlst))
		;; since reruns have been tacked on to newlst create new reruns from junked
		(loop (car newlst)(cdr newlst) reg (delete-duplicates junked)))))
	 ((not (null? tal))
          (debug:print-info 4 *default-log-port* "cond branch - "  "rtq-7")
	  (debug:print-info 4 *default-log-port* "I'm pretty sure I shouldn't get here."))
	 ((not (null? reg)) ;; could we get here with leftovers?
          (debug:print-info 4 *default-log-port* "cond branch - "  "rtq-8")
	  (debug:print-info 0 *default-log-port* "Have leftovers!")
	  (loop (car reg)(cdr reg) '() reruns))
	 (else
          (debug:print-info 4 *default-log-port* "cond branch - "  "rtq-9")
	  (debug:print-info 4 *default-log-port* "Exiting loop with...\n  hed=" hed "\n  tal=" tal "\n  reruns=" reruns))
	 ))) ;; end loop on sorted test names
    ;; this is the point where everything is launched and now you can mark the run in metadata table as all launched 
    (rmt:set-var run-id (conc "lunch-complete-" run-id) "yes")  
        
    ;; now *if* -run-wait we wait for all tests to be done
    ;; Now wait for any RUNNING tests to complete (if in run-wait mode)
    ;; (if (runs:dat-load-mgmt-function runsdat)((runs:dat-load-mgmt-function runsdat)))
    (thread-sleep! 10) ;; I think there is a race condition here. Let states/statuses settle
    
    (let wait-loop ((num-running      (rmt:get-count-tests-running-for-run-id run-id))
		    (prev-num-running 0))
      ;; (debug:print-info 13 *default-log-port* "num-running=" num-running ", prev-num-running=" prev-num-running)
      (if (and (or (args:get-arg "-run-wait")
		   (equal? (configf:lookup *configdat* "setup" "run-wait") "yes"))
	       (> num-running 0))
	  (begin
	    ;; Here we mark any old defunct tests as incomplete. Do this every fifteen minutes
	    ;; (debug:print 0 *default-log-port* "Got here eh! num-running=" num-running " (> num-running 0) " (> num-running 0))
	    (if (> (current-seconds)(+ last-time-incomplete 900))
		(let ((actual-num-running (rmt:get-count-tests-running-for-run-id run-id)))
		  (debug:print-info 0 *default-log-port* "Marking stuck tests as INCOMPLETE while waiting for run " run-id
				    ". Running as pid " (current-process-id) " on " (get-host-name))
		  (set! last-time-incomplete (current-seconds)) ;; FIXME, this might be causing slow down - use of set!
		  (runs:find-and-mark-incomplete-and-check-end-of-run run-id #f)
		  (debug:print-info 0 *default-log-port* "run-wait specified, waiting on " actual-num-running
				    " tests in RUNNING, REMOTEHOSTSTART or LAUNCHED state at "
				    (time->string (seconds->local-time (current-seconds))))))
	    ;; (if (runs:dat-load-mgmt-function runsdat)((runs:dat-load-mgmt-function runsdat)))
	    (thread-sleep! 5) ;; (if (>= num-running max-concurrent-jobs) 5 1))
	    (wait-loop (rmt:get-count-tests-running-for-run-id run-id)
		       num-running))))
    ;; LET* ((test-record
    ;; we get here on "drop through". All done!
    ;; this is moved to runs:run-testes since this function is getting called twice to ensure everthing is completed. 
    ;; (debug:print-info 0 *default-log-port* "Calling Post Hook")    
    ;; (runs:run-post-hook run-id)
    (debug:print-info 1 *default-log-port* "All tests launched")))

(define (runs:calc-fails prereqs-not-met)
  (filter (lambda (test)
	    (and (vector? test) ;; not (string? test))
		 (member (db:test-get-state test) '("INCOMPLETE" "COMPLETED")) ;; TODO: pull from *common:stuff...*
		 (not (member (db:test-get-status test)
			      '("PASS" "WARN" "WAIVED" "SKIP")))))
	  prereqs-not-met))

(define (runs:calc-prereq-fail prereqs-not-met) ;; REMOVEME since NOT_STARTED/PREQ_FAIL is now COMPLETED/PREQ_FAIL
  (filter (lambda (test)
	    (and (vector? test) ;; not (string? test))
		 (equal? (db:test-get-state test) "NOT_STARTED")
		 (not (member (db:test-get-status test)
			      '("n/a" "KEEP_TRYING")))))
	  prereqs-not-met))

(define (runs:calc-not-completed prereqs-not-met)
  (filter
   (lambda (t)
     (or (not (vector? t))
	 (not (member (db:test-get-state t) '("INCOMPLETE" "COMPLETED")))))
   prereqs-not-met))

;; (define (runs:calc-not-completed prereqs-not-met)
;;   (filter
;;    (lambda (t)
;;      (or (not (vector? t))
;; 	 (not (equal? "COMPLETED" (db:test-get-state t)))))
;;    prereqs-not-met))

(define (runs:calc-runnable prereqs-not-met)
  (filter 
   (lambda (t)
     (or (not (vector? t))
	 (and (equal? "NOT_STARTED" (db:test-get-state t))
	      (member (db:test-get-status t)
		      '("n/a" "KEEP_TRYING")))
	 (and (equal? "RUNNING" (db:test-get-state t))))) ;; account for a test that is running
   prereqs-not-met))

(define (runs:pretty-string lst)
  (map (lambda (t)
	 (if (not (vector? t))
	     (conc t)
	     (conc (db:test-get-testname t) ":" (db:test-get-state t) "/" (db:test-get-status t))))
       lst))

;; parent-test is there as a placeholder for when parent-tests can be run as a setup step
;;
(define (run:test run-id run-info keyvals runname test-record flags parent-test test-registry all-tests-registry runsdat testdat-rec)
  ;; All these vars might be referenced by the testconfig file reader
  ;; 
  ;; NEED to reprocess testconfig here, ensuring that item variables are available.
  ;; This is for Tal's issue with item-specific env vars not being set for use in skip.
  ;; HSD https://hsdes.intel.com/appstore/icf/index.html#/article?articleId=1408763273
  ;;
  (let* ((test-name    (tests:testqueue-get-testname   test-record))
	 (test-waitons (tests:testqueue-get-waitons    test-record))
	 (itemdat      (tests:testqueue-get-itemdat    test-record))
	 (item-path     "")
	 (db           #f)
	 (full-test-name #f))

    ;; setting itemdat to a list if it is #f
    (if (not itemdat)(set! itemdat '()))
    (set! item-path (item-list->path itemdat))
    (set! full-test-name (db:test-make-full-name test-name item-path))
    (runs:set-megatest-env-vars run-id inrunname: runname testname: test-name itempath: item-path) ;; these may be needed by the launching process

   (let* ((test-conf ;; re-instate the tests:get-testconfig once the kinks are worked out. FIXME!!!
	    ;; (tests:get-testconfig test-name item-path all-tests-registry #t force-create: #t))
	    (tests:testqueue-get-testconfig test-record )) 
	   (test-path    (hash-table-ref all-tests-registry test-name)) ;; (conc *toppath* "/tests/" test-name)) ;; could use tests:get-testconfig here ...

	   (force        (hash-table-ref/default flags "-force" #f))
	   (rerun        (hash-table-ref/default flags "-rerun" #f))
	   (keepgoing    (hash-table-ref/default flags "-keepgoing" #f))
	   (incomplete-timeout (string->number (or (configf:lookup *configdat* "setup" "incomplete-timeout") "x")))
	   )

    (debug:print-info 4 *default-log-port*
		      "\nTESTNAME: " full-test-name 
		      "\n   test-config: " (hash-table->alist test-conf)
		      "\n   itemdat: " itemdat
		      )
    (debug:print 2 *default-log-port* "Attempting to launch test " full-test-name)
    ;; (set-environment-variable! "MT_TEST_NAME" test-name) ;; 
    ;; (set-environment-variable! "MT_ITEMPATH"  item-path)
    ;; (set-environment-variable! "MT_RUNNAME"   runname)
    (change-directory *toppath*)

    ;; Here is where the test_meta table is best updated
    ;; Yes, another use of a global for caching. Need a better way?
    ;;
    ;; There is now a single call to runs:update-all-test_meta and this 
    ;; per-test call is not needed. Given the delicacy of the move to 
    ;; v1.55 this code is being left in place for the time being.
    ;;
    (if (not (hash-table-ref/default *test-meta-updated* test-name #f))
        (begin
          (hash-table-set! *test-meta-updated* test-name #t)
          (runs:update-test_meta test-name test-conf)))
    
    ;; itemdat => ((ripeness "overripe") (temperature "cool") (season "summer"))
    (let* ((new-test-path (string-intersperse (cons test-path (map cadr itemdat)) "/"))
	   (test-id       (rmt:get-test-id run-id test-name item-path))
	   (testdat       (if test-id (rmt:get-test-info-by-id run-id test-id) #f)))
      (if (not testdat)
	  (let loop ()
	    ;; ensure that the path exists before registering the test
	    ;; NOPE: Cannot! Don't know yet which disk area will be assigned....
	    ;; (system (conc "mkdir -p " new-test-path))
	    ;;
	    ;; (open-run-close tests:register-test db run-id test-name item-path)
	    ;;
	    ;; NB// for the above line. I want the test to be registered long before this routine gets called!
	    ;;
	    (if (not test-id)(set! test-id (rmt:get-test-id run-id test-name item-path)))
	    (if (not test-id)
		(begin
		  (debug:print 2 *default-log-port* "WARN: Test not pre-created? test-name=" test-name ", item-path=" item-path ", run-id=" run-id)
		  (rmt:register-test run-id test-name item-path)
		  (set! test-id (rmt:get-test-id run-id test-name item-path))))
	    (debug:print-info 4 *default-log-port* "test-id=" test-id ", run-id=" run-id ", test-name=" test-name ", item-path=\"" item-path "\"")
	    (set! testdat (rmt:get-test-info-by-id run-id test-id))
	    (if (not testdat)
		(begin
		  (debug:print-info 0 *default-log-port* "WARNING: server is overloaded, trying again in two seconds")
		  ;; (if (runs:dat-load-mgmt-function runsdat)((runs:dat-load-mgmt-function runsdat)))
		  (thread-sleep! 2)
		  (loop)))))
      (if (not testdat) ;; should NOT happen
	  (debug:print-error 0 *default-log-port* "failed to get test record for test-id " test-id))
      (set! test-id (db:test-get-id testdat))
      (if (common:file-exists? test-path)
	  (change-directory test-path)
	  (begin
	    (debug:print-error 0 *default-log-port* "test run path not created before attempting to run the test. Perhaps you are running -remove-runs at the same time?")
	    (change-directory *toppath*)))
      (case (if force ;; (args:get-arg "-force")
		'NOT_STARTED
		(if testdat
		    (string->symbol (test:get-state testdat))
		    'failed-to-insert))
	((failed-to-insert)
	 (debug:print-error 0 *default-log-port* "Failed to insert the record into the db"))
	((NOT_STARTED COMPLETED DELETED INCOMPLETE)
	 (let ((runflag #f))
	   (cond
	    ;; -force, run no matter what
	    (force (set! runflag #t))
	    ;; NOT_STARTED, run no matter what
	    ((member (test:get-state testdat) '("DELETED" "NOT_STARTED" "INCOMPLETE"))(set! runflag #t))
	    ;; not -rerun and PASS, WARN or CHECK, do no run
	    ((and (or (not rerun)
		      keepgoing)
		  ;; Require to force re-run for COMPLETED or *anything* + PASS,WARN or CHECK
		  (or (member (test:get-status testdat) '("PASS" "WARN" "CHECK" "SKIP" "WAIVED"))
		      (member (test:get-state  testdat) '("COMPLETED")))) 
	     (debug:print-info 2 *default-log-port* "running test " test-name "/" item-path " suppressed as it is " (test:get-state testdat) " and " (test:get-status testdat))
	     (hash-table-set! test-registry full-test-name 'DONOTRUN) ;; COMPLETED)
	     (set! runflag #f))

	    ;; -rerun and status is one of the specifed, run it
	    ((and rerun
		  (let* ((rerunlst   (string-split rerun ","))
			 (must-rerun (member (test:get-status testdat) rerunlst)))
		    (debug:print-info 3 *default-log-port* "-rerun list: " rerun ", test-status: " (test:get-status testdat)", must-rerun: " must-rerun)
		    must-rerun))
	     (debug:print-info 2 *default-log-port* "Rerun forced for test " test-name "/" item-path)
	     (set! runflag #t)
             (debug:print-info 2 *default-log-port* "Calling rerun hook")
             (runs:rerun-hook test-id new-test-path testdat rerun)
             )
             

	    
            ;; -keepgoing, do not rerun FAIL
	    ((and keepgoing
		  (member (test:get-status testdat) '("FAIL")))
	     (set! runflag #f))
	    
            ((and (not rerun)
		  (member (test:get-status testdat) '("FAIL" "n/a")))
	     (set! runflag #t))

	    (else (set! runflag #f)))
	   (debug:print 4 *default-log-port* "RUNNING => runflag: " runflag " STATE: " (test:get-state testdat) " STATUS: " (test:get-status testdat))
	   (if (not runflag)
	       (if (not parent-test)
		   (if (runs:lownoise (conc "not starting test" full-test-name) 60)
		       (debug:print 1 *default-log-port* "NOTE: Not starting test " full-test-name " as it is state \"" (test:get-state testdat) 
				    "\" and status \"" (test:get-status testdat) "\", use -rerun \"" (test:get-status testdat)
				    "\" or -force to override")))
	       ;; NOTE: No longer be checking prerequisites here! Will never get here unless prereqs are
	       ;;       already met.
	       ;; This would be a great place to do the process-fork
	       ;; 
	       (let ((skip-test   #f)
		     (skip-check  (configf:get-section test-conf "skip")))
		 (cond 
		  ;; Have to check for skip conditions. This one skips if there are same-named tests
		  ;; currently running
		  ((and skip-check
			(configf:lookup test-conf "skip" "prevrunning"))
		   ;; run-ids = #f means *all* runs
		   (let ((running-tests (rmt:get-tests-for-runs-mindata #f full-test-name '("RUNNING" "REMOTEHOSTSTART" "LAUNCHED") '() #f)))
		     (if (not (null? running-tests)) ;; have to skip 
			 (set! skip-test "Skipping due to previous tests running"))))

 		    ;; split the string and OR of file-exists?
		    ((and skip-check
			  (configf:lookup test-conf "skip" "fileexists"))
		     (let* ((files    (string-split (configf:lookup test-conf "skip" "fileexists")))
			    (existing (filter common:file-exists? files)))
		       (if (not (null? existing)) ;; (common:file-exists? (configf:lookup test-conf "skip" "fileexists"))
			   (set! skip-test (conc "Skipping due to existance of file(s) " (string-intersperse existing ", ")))))) ;; (configf:lookup test-conf "skip" "fileexists")))))

 		    ((and skip-check
			  (configf:lookup test-conf "skip" "filenotexists"))
		      (let* ((files    (string-split (configf:lookup test-conf "skip" "filenotexists")))
			      (existing (filter common:file-exists? files)))
			      (if (null? existing) ;;  (common:file-exists? (configf:lookup test-conf "skip" "filenotexists")))
			        (set! skip-test (conc "Skipping due to non existance of files " (string-intersperse files ", ")))))) ;; (configf:lookup test-conf "skip" "filenotexists")))))        
		 
       ((and skip-check
			(configf:lookup test-conf "skip" "script"))
		   (if (= (system (configf:lookup test-conf "skip" "script")) 0)
                        (set! skip-test (conc "Skipping due to zero return value of script " (configf:lookup test-conf "skip" "script")))))
      
                  ((and skip-check
			(configf:lookup test-conf "skip" "rundelay"))
		   ;; run-ids = #f means *all* runs
		   (let* ((numseconds      (common:hms-string->seconds (configf:lookup test-conf "skip" "rundelay")))
			  (running-tests   (rmt:get-tests-for-runs-mindata #f full-test-name '("RUNNING" "REMOTEHOSTSTART" "LAUNCHED") '() #f))
			  (completed-tests (rmt:get-tests-for-runs-mindata #f full-test-name '("COMPLETED" "INCOMPLETE") '("PASS" "FAIL" "ABORT") #f)) ;; ironically INCOMPLETE is same as COMPLETED in this contex
			  (last-run-times  (map db:mintest-get-event_time completed-tests))
			  (time-since-last (- (current-seconds) (if (null? last-run-times) 0 (common:max last-run-times)))))
		     (if (or (not (null? running-tests)) ;; have to skip if test is running
			     (> numseconds time-since-last))
			 (set! skip-test (conc "Skipping due to previous test run less than " (configf:lookup test-conf "skip" "rundelay") " ago"))))))
		 
		 (if skip-test
		     (begin
		       (mt:test-set-state-status-by-id run-id test-id "COMPLETED" "SKIP" skip-test)
		       (debug:print-info 1 *default-log-port* "SKIPPING Test " full-test-name " due to " skip-test))
		     ;;
		     ;; Here the test is handed off to launch.scm for launch-test to complete the launch process
		     ;;
		     (begin
		       ;; wait for less than max jobs here
		       (if (runs:dat-wait-for-jobs-function runsdat)
			   ((runs:dat-wait-for-jobs-function runsdat) testdat-rec))
		       
		       (if (not (launch-test test-id run-id run-info keyvals runname test-conf test-name test-path itemdat flags))
			   (begin
			     (print "ERROR: Failed to launch the test. Exiting as soon as possible")
			     (set! *globalexitstatus* 1) ;; 
			     (process-signal (current-process-id) signal/kill))
			   )
		       ;; wait again here?
		       ))))))
	((KILLED) 
	 (debug:print 1 *default-log-port* "NOTE: " full-test-name " is already running or was explictly killed, use -force to launch it.")
	 (hash-table-set! test-registry (db:test-make-full-name test-name test-path) 'DONOTRUN)) ;; KILLED))
	((LAUNCHED REMOTEHOSTSTART RUNNING)  
	 (debug:print 2 *default-log-port* "NOTE: " test-name " is already running"))
	;; (if (> (- (current-seconds)(+ (db:test-get-event_time testdat)
	;; 			       (db:test-get-run_duration testdat)))
	;; 	(or incomplete-timeout
	;; 	    6000)) ;; i.e. no update for more than 6000 seconds
	;;      (begin
	;;        (debug:print 0 *default-log-port* "WARNING: Test " test-name " appears to be dead. Forcing it to state INCOMPLETE and status STUCK/DEAD")
	;;        (tests:test-set-status! run-id test-id "INCOMPLETE" "STUCK/DEAD" "" #f))
	;;        ;; (tests:test-set-status! test-id "INCOMPLETE" "STUCK/DEAD" "" #f))
	;;      (debug:print 2 *default-log-port* "NOTE: " test-name " is already running")))
	(else      
	 (debug:print-error 0 *default-log-port* "Failed to launch test " full-test-name ". Unrecognised state " (test:get-state testdat))
	 (case (string->symbol (test:get-state testdat)) 
	   ((COMPLETED INCOMPLETE)
	    (hash-table-set! test-registry (db:test-make-full-name test-name test-path) 'DONOTRUN))
	   (else
	    (hash-table-set! test-registry (db:test-make-full-name test-name test-path) 'DONOTRUN)))))))))

;;======================================================================
;; END OF NEW STUFF
;;======================================================================

(define (get-dir-up-n dir . params) 
  (let ((dparts  (string-split dir "/"))
	(count   (if (null? params) 1 (car params))))
    (conc "/" (string-intersperse 
	       (take dparts (- (length dparts) count))
	       "/"))))

;; cleanup often needs to remove all but the last N runs per target
;;
;; target-patts a1/b1/c1,a2/b2/c2 ...
;;
;; This will fail if called with empty target or a bad target (i.e. missing or extra fields)
;;
(define (runs:get-hash-by-target target-patts runpatt)
  (let* ((targets (string-split target-patts ","))
	 (keys    (rmt:get-keys))
	 (res-ht  (make-hash-table))) ;; target -> ( runrecord1 runrecord2 ... )
    (for-each
     (lambda (target-patt)
       (let ((runs     (rmt:simple-get-runs runpatt #f #f target-patt #f)))
	 (for-each
	  (lambda (run)
	    (let ((target (simple-run-target run)))
	      (hash-table-set! res-ht target (cons run (hash-table-ref/default res-ht target '())))))
	  runs)))
     targets)
    res-ht))

;; delete runs older than X (weeks, days, months years etc.)
;; delete redundant runs within a target - N is the input
;; delete redundant runs within a target IFF older than given date/time AND keep at least N
;; 
(define (runs:remove-all-but-last-n-runs-per-target target-patts runpatt num-to-keep #!key (actions '(print)))
  (let* ((runs-ht  (runs:get-hash-by-target target-patts runpatt))
	 (age      (if (args:get-arg "-age")(common:hms-string->seconds (args:get-arg "-age")) #f))
	 (age-mark (if age (- (current-seconds) age) (+ (current-seconds) 86400)))
	 (precmd   (or (args:get-arg "-precmd") ""))
         (action-chk (member (string->symbol "remove-runs") actions)))
     ;; check the sequence of actions  archive must comme before remove-runs
     (if  (and action-chk (member (string->symbol "archive") action-chk))
          (begin
          (debug:print-error 0 *default-log-port* "action remove-runs must come after archive")
          (exit 1))) 
    (print "Actions: " actions " age: " age)
    (for-each
     (lambda (action)
        (for-each
         (lambda (target)
            (let* ((runs      (hash-table-ref runs-ht target))
	           (sorted    (sort runs (lambda (a b)(< (simple-run-event_time a)(simple-run-event_time b)))))
	           (to-remove (let* ((len      (length sorted))
                                     (trim-amt (- len num-to-keep)))
                                 (if (> trim-amt 0)
                                    (take sorted trim-amt)
                                    '()))))
	   (hash-table-set! runs-ht target to-remove)
           (print target ":")
           (for-each
            (lambda (run)
              (let ((remove (member run to-remove (lambda (a b)
                                                    (eq? (simple-run-id a)
                                                         (simple-run-id b))))))
  	        (if (and age (> (simple-run-event_time run) age-mark))
		     (print "Skipping handling of " target "/" (simple-run-runname run) " as it is younger than " (args:get-arg "-age"))
		     (case action
		       ((print)
			(print " " (simple-run-runname run)
			       " " (time->string (seconds->local-time (simple-run-event_time run)) "WW%V.%u %H:%M:%S")
			       " " (if remove "REMOVE" "")))
		       ((remove-runs)
			(if remove (system (conc precmd " megatest -remove-runs -target " target " -runname " (simple-run-runname run) " -testpatt %"
						 (if (member 'kill-runs actions) ;; if kill-runs is specified then set -kill-wait to 0
						     " -kill-wait 0"
						     "")))))
		       ((archive)
			(if remove (system (conc precmd " megatest -archive save-remove -target " target " -runname " (simple-run-runname run) " -testpatt %"))))
		       ((kill-runs)
			(if remove (system (conc precmd " megatest -kill-runs -target " target " -runname " (simple-run-runname run) " -testpatt %"))))))))
            sorted)))
         (hash-table-keys runs-ht)))
      actions)
    runs-ht))

(define (remove-last-path-directory path-in)
  (let* ((dparts  (string-split path-in "/"))
    (path-out (conc "/" (string-intersperse (take dparts (- (length dparts) 1)) "/")))
    )
    path-out
  )
)

(define (runs:operate-on action target runnamepatt testpatt #!key (state #f)(status #f)(new-state-status #f)(mode #f)(options '()))
  (common:clear-caches) ;; clear all caches
  (let* ((db           #f)
	 ;; (tdbdat       (tasks:open-db))
	 (keys         (rmt:get-keys))
	 (rundat       (mt:get-runs-by-patt keys runnamepatt target))
	 (header       (vector-ref rundat 0))
	 (runs         (vector-ref rundat 1))
	 (states       (if state  (string-split state  ",") '()))
	 (statuses     (if status (string-split status ",") '()))
	 (state-status (if (string? new-state-status) (string-split new-state-status ",") '(#f #f)))
	 (rp-mutex     (make-mutex))
	 (bup-mutex    (make-mutex))
         (keep-records (args:get-arg "-keep-records")) ;; used in conjunction with -remove-runs to keep the records, TODO: consolidate this with "mode".
	 (test-records '())) ;; for tasks that we wish to operate on all tests in one fell swoop

    (let* ((write-access-actions '(remove-runs set-state-status archive run-wait kill-runs))
           (dbfile             (conc  *toppath* "/.db/main.db"))
           (readonly-mode      (not (file-writable? dbfile))))
      (when (and readonly-mode
                 (member action write-access-actions))
        (debug:print-error 0 *default-log-port* ".db/main.db is readonly.  Cannot proceed with action ["action"] in which write-access isrequired .")
        (exit 1)))
    
    (debug:print-info 4 *default-log-port* "runs:operate-on => Header: " header " action: " action " new-state-status: " new-state-status)
    (if (> 2 (length state-status))
	(begin
	  (debug:print-error 0 *default-log-port* "the parameter to -set-state-status is a comma delimited string. E.g. COMPLETED,FAIL")
	  (exit)))
    (for-each
     (lambda (run)
       (let ((runkey (string-intersperse (map (lambda (k)
						(db:get-value-by-header run header k)) keys) "/"))
	     (dirs-to-remove (make-hash-table))
	     (proc-get-tests (lambda (run-id)
                               (mt:get-tests-for-run run-id
                                                     testpatt states statuses
                                                     not-in:  #f
                                                     sort-by: (case action
                                                                ((remove-runs) 'rundir)
                                                                (else          'event_time))))))
	 (let* ((run-id    (db:get-value-by-header run header "id"))
		(run-state (db:get-value-by-header run header "state"))
		(run-name  (db:get-value-by-header run header "runname"))
		(tests     (if (not (equal? run-state "locked"))
			       (proc-get-tests run-id)
			       '()))
		(lasttpath "/does/not/exist/I/hope")
		(lastrealpath "/does/not/exist/I/hope")
                ;; there may be a number of different disks used in the same run.
                (run-paths-hash (make-hash-table))
		(worker-thread #f))
	   (debug:print-info 4 *default-log-port* "runs:operate-on run=" run ", header=" header)
	   (if (not (null? tests))
	       (begin
		 (case action
                   ((kill-runs)
                    (tasks:kill-runner target run-name "%")
                    (debug:print 1 *default-log-port* "Killing tests for run: " runkey " " (db:get-value-by-header run header "runname"))
                    )
		   ((remove-runs)
		    ;; (if (tasks:need-server run-id)(tasks:start-and-wait-for-server tdbdat run-id 10))
		    ;; seek and kill in flight -runtests with % as testpatt here
		    ;; (if (equal? testpatt "%")
		    (tasks:kill-runner target run-name testpatt)
		    ;; (debug:print 0 *default-log-port* "not attempting to kill any run launcher processes as testpatt is " testpatt))
		    (debug:print 1 *default-log-port* "Removing tests for run: " runkey " " (db:get-value-by-header run header "runname")))
		   ((set-state-status)
		    ;; (if (tasks:need-server run-id)(tasks:start-and-wait-for-server tdbdat run-id 10))
		    (debug:print 1 *default-log-port* "Modifying state and staus for tests for run: " runkey " " (db:get-value-by-header run header "runname")))
		   ((print-run)
		    (debug:print 1 *default-log-port* "Printing info for run " runkey ", run=" run ", tests=" tests ", header=" header)
		    action)
		   ((run-wait)
		    (debug:print 1 *default-log-port* "Waiting for run " runkey ", run=" runnamepatt " to complete"))
		   ((archive)
		    (debug:print 1 *default-log-port* "Archiving/restoring (" (args:get-arg "-archive") ") data for run: " runkey " " (db:get-value-by-header run header "runname"))
		    (let ((op (string->symbol (args:get-arg "-archive"))))
		      (set! worker-thread
			    (make-thread
			     (lambda ()
			       (case op
				 ((save save-remove keep-html)
				  (archive:run-bup op run-id run-name tests rp-mutex bup-mutex))
				 ((restore)
				  (archive:bup-restore op run-id run-name tests rp-mutex bup-mutex))
				 ((get) ;;; NOTE: This is a special case. We wish to operate on ALL tests in one go
				  (set! test-records (append tests test-records)))
				 (else 
				  (debug:print-error 0 *default-log-port* "unrecognised sub command " op " for -archive. Run \"megatest\" to see help")
				  (exit))))
			     "archive-bup-thread"))
		      (thread-start! worker-thread)
		      (if (eq? op 'get)
			  (thread-join! worker-thread)) ;; we need the test-records set to not overlap
		      ))
		   (else
		    (debug:print-info 0 *default-log-port* "action not recognised " action)))
		 
		 ;; actions that operate on one test at a time can be handled below
		 ;;
		 (let ((sorted-tests     (filter 
					  vector?
					  (sort tests (lambda (a b)(let ((dira ;; (rmt:sdb-qry 'getstr 
									  (db:test-get-rundir a)) ;; )  ;; (filedb:get-path *fdb* (db:test-get-rundir a)))
									 (dirb ;; (rmt:sdb-qry 'getstr 
									  (db:test-get-rundir b))) ;; ) ;; ((filedb:get-path *fdb* (db:test-get-rundir b))))
								     (if (and (string? dira)(string? dirb))
									 (> (string-length dira)(string-length dirb))
									 #f))))))
		       (toplevel-retries (make-hash-table)) ;; try three times to loop through and remove top level tests
		       (test-retry-time  (make-hash-table))
                       (backgrounded-remove-status     (make-hash-table))
                       (backgrounded-remove-last-visit (make-hash-table))
                       (backgrounded-remove-result     (make-hash-table))
		       (allow-run-time   (string->number (or (args:get-arg "-kill-wait") "10")))) ;; seconds to allow for killing tests before just brutally killing 'em
		   (let loop ((test (car sorted-tests))
			      (tal  (cdr sorted-tests)))
		     (let* ((test-id       (db:test-get-id test))
			    (new-test-dat  (rmt:get-test-info-by-id run-id test-id)))
		       (if (not new-test-dat)
			   (begin
			     (debug:print-error 0 *default-log-port* "We have a test-id of " test-id " but no record was found. NOTE: No locking of records is done between processes, do not simultaneously remove the same run from two processes!")
			     (if (not (null? tal))
				 (loop (car tal)(cdr tal))))
			   (let* ((item-path     (db:test-get-item-path new-test-dat))
				  (test-name     (db:test-get-testname new-test-dat))
				  (run-dir       ;;(filedb:get-path *fdb*
				   ;; (rmt:sdb-qry 'getid 
				   (db:test-get-rundir new-test-dat)) ;; )    ;; run dir is from the link tree
                                  (has-subrun    (and (subrun:subrun-test-initialized? run-dir)
                                                      (not (subrun:subrun-removed? run-dir))))
				  (test-state    (db:test-get-state new-test-dat))
                                  (test-status    (db:test-get-status new-test-dat))
				  (test-fulln    (db:test-get-fullname new-test-dat))
				  (uname         (db:test-get-uname    new-test-dat))
				  (toplevel-with-children (and (db:test-get-is-toplevel test)
							       (> (rmt:test-toplevel-num-items run-id test-name) 0))))
                             
			     (case action
			       ((remove-runs)
				;; if the test is a toplevel-with-children issue an error and do not remove
				(cond
                                 (toplevel-with-children
                                  (debug:print 0 *default-log-port* "WARNING: skipping removal of " test-fulln " with run-id " run-id " as it has sub tests")
                                  (hash-table-set! toplevel-retries test-fulln (+ (hash-table-ref/default toplevel-retries test-fulln 0) 1))
                                  (if (> (hash-table-ref toplevel-retries test-fulln) 3)
                                      (if (not (null? tal))
                                          (loop (car tal)(cdr tal))) ;; no else clause - drop it if no more in queue and > 3 tries
                                      (let ((newtal (append tal (list test))))
                                        (loop (car newtal)(cdr newtal))))) ;; loop with test still in queue
                                 (has-subrun
                                  ;; 
                                  (let ((last-visit (hash-table-ref/default backgrounded-remove-last-visit test-fulln 0))
                                        (now        (current-seconds))
                                        (rem-status (hash-table-ref/default backgrounded-remove-status test-fulln 'not-started)))
                                    (case rem-status
                                      ((not-started)
                                       (debug:print 0 *default-log-port* "WARNING: postponing removal of " test-fulln " with run-id " run-id " as it has a subrun")
                                       (hash-table-set! backgrounded-remove-status test-fulln 'started)
                                       (hash-table-set! backgrounded-remove-last-visit test-fulln (current-seconds))
                                       (common:send-thunk-to-background-thread
                                        (lambda ()
                                          (let* ((subrun-remove-succeeded
                                                  (subrun:remove-subrun run-dir keep-records)))
                                            (hash-table-set! backgrounded-remove-result test-fulln subrun-remove-succeeded)
                                            (hash-table-set! backgrounded-remove-status test-fulln 'done)))
                                        name: (conc "remove-subrun:"test-fulln))
                                       
                                       ;; send to back of line, loop
                                       (let ((newtal (append tal (list test))))
                                        (loop (car newtal)(cdr newtal)))
                                       )
                                      ((started)
                                       ;; if last visit was within last second, sleep 1 second
                                       (if (< (- now last-visit) 1.0)
                                           (thread-sleep! 1.0))
                                       (hash-table-set! backgrounded-remove-last-visit test-fulln (current-seconds))
                                       ;; send to back of line, loop
                                       (let ((newtal (append tal (list test))))
                                         (loop (car newtal)(cdr newtal)))
                                       )
                                      ((done)
                                       ;; drop this one; if remaining, loop, else finish
                                       (hash-table-set! backgrounded-remove-last-visit test-fulln (current-seconds))
                                       (let ((subrun-remove-succeeded (hash-table-ref/default backgrounded-remove-result test-fulln 'exception)))
                                         (cond
                                          ((eq? subrun-remove-succeeded 'exception)
                                           (let* ((logfile (subrun:get-log-path run-dir "remove")))
                                             (debug:print 0 *default-log-port* "ERROR: removing subrun of of " test-fulln " with run-id " run-id " ; see logfile @ "logfile))
					   (if (not (null? tal))
					       (loop (car tal)(cdr tal))))
					  (subrun-remove-succeeded
					   (debug:print 0 *default-log-port* "Now removing of " test-fulln " with run-id " run-id " since subrun was removed.")
					   ;;(runs:remove-test-directory new-test-dat mode) ;; let normal case handle this. it will go thru loop again as non-subrun
					   (let ((newtal (append tal (list test))))
					     (loop (car newtal)(cdr newtal))))
					  (else
					   (let* ((logfile (subrun:get-log-path run-dir "remove")))
					     (debug:print 0 *default-log-port* "WARNING: removal of subrun failed.  Please check "logfile" for details."))
					   ;; send to back of line, loop (will not match has-subrun next time through)
					    (if (not (null? tal))
						(loop (car tal)(cdr tal))))))
				       )
				      ) ; end case rem-status
                                    ) ; end let
                                  ); end cond has-subrun

                                 (else
                                  ;; BB - TODO - consider backgrounding to threads to delete tests (work below) 
                                  (debug:print-info 0 *default-log-port* "test: " test-name " itest-state: " test-state)
                                  (if (member test-state (list "RUNNING" "LAUNCHED" "REMOTEHOSTSTART" "KILLREQ"))
                                      (begin
                                        (if (not (hash-table-ref/default test-retry-time test-fulln #f))
                                            (begin
                                              ;; want to set to REMOVING BUT CANNOT do it here?
                                              (hash-table-set! test-retry-time test-fulln (current-seconds))))
                                        (if (> (- (current-seconds)(hash-table-ref test-retry-time test-fulln)) allow-run-time)
                                            ;; This test is not in a correct state for cleaning up. Let's try some graceful shutdown steps first
                                            ;; Set the test to "KILLREQ" and wait five seconds then try again. Repeat up to five times then give
                                            ;; up and blow it away.
                                            (begin
                                              (debug:print 0 *default-log-port* "WARNING: could not gracefully remove test " test-fulln ", tried to kill it to no avail. Forcing state to FAILEDKILL and continuing")
                                              (mt:test-set-state-status-by-id run-id (db:test-get-id test) "FAILEDKILL" "n/a" #f)
                                              (thread-sleep! 1))
                                            (begin
                                              (mt:test-set-state-status-by-id run-id (db:test-get-id test) "KILLREQ" "n/a" #f)
                                              (thread-sleep! 1)))
                                        ;; NOTE: This is suboptimal as the testdata will be used later and the state/status may have changed ...
                                        (if (null? tal)
                                            (loop new-test-dat tal)
                                            (loop (car tal)(append tal (list new-test-dat)))))
                                      (begin
                                       (let ((rundir (db:test-get-rundir new-test-dat)))
                                        (if (and (not (string=  rundir "/tmp/badname")) 
                                             (file-exists? rundir)
                                             (substring-index run-name rundir)
                                             (tests:glob-like-match (conc "%/" target "/%") rundir)
                                             )
                                          (begin
                                            (set! lasttpath (db:test-get-rundir new-test-dat)) ;; remember this path for run removal
                                            (set! lastrealpath (remove-last-path-directory (realpath lasttpath)))
                                            (hash-table-set! run-paths-hash lastrealpath 1)
                                            (runs:remove-test-directory new-test-dat mode) ;; 'remove-all)
                                          )
                                          (begin
                                            (debug:print 2 *default-log-port* "Not removing directory " rundir " because either it doesn't exist or has a bad name")
                                            (debug:print 2 *default-log-port* "Is /tmp/badname: " (string=  rundir "/tmp/badname"))
                                            (debug:print 2 *default-log-port* "Exists: " (file-exists? rundir))
                                            (debug:print 2 *default-log-port* "Has run-name: " (substring-index run-name rundir))
                                            (debug:print 2 *default-log-port* "Has target: " (tests:glob-like-match (conc "%/" target "/%") rundir))
                                            (debug:print 2 *default-log-port* "Target: " target)

                                            ;;PJH remove record from db no need to cleanup directory
                                            (case mode
                                               ((remove-data-only)(mt:test-set-state-status-by-id (db:test-get-run_id test)(db:test-get-id test) (db:test-get-state test)(db:test-get-status test) #f))
                                               ((archive-remove)  (mt:test-set-state-status-by-id (db:test-get-run_id test)(db:test-get-id test) "ARCHIVED" #f #f))
                                               (else (rmt:delete-test-records (db:test-get-run_id test) (db:test-get-id test))))

                                          )
                                        )
                                      )

                                      (if (not (null? tal))
                                         (loop (car tal)(cdr tal)))))))
				(rmt:update-run-stats run-id (rmt:get-raw-run-stats run-id)))
                               ((kill-runs)
                                ;; RUNNING -> KILLREQ
                                ;; LAUNCHED,RUNNING,REMOTEHOSTSTART -> NOT STARTED
                                (cond
                                 ((and has-subrun (member test-state (list "RUNNING" "LAUNCHED" "REMOTEHOSTSTART" "KILLREQ")))
                                  (common:send-thunk-to-background-thread
                                   (lambda ()
                                     (let* ((subrun-remove-succeeded
                                             (subrun:kill-subrun run-dir keep-records)))
                                       #t)))
                                  (if (not (null? tal))
				    (loop (car tal)(cdr tal)))
                                  )
                                 ((member test-state (list "RUNNING" "LAUNCHED" "REMOTEHOSTSTART" "KILLREQ"))
                                  (debug:print 1 *default-log-port* "INFO: issuing killreq to test "test-fulln)
                                  (mt:test-set-state-status-by-id run-id (db:test-get-id test) "KILLREQ" "n/a" #f)
                                  (if (not (null? tal))
				      (loop (car tal)(cdr tal))))
                                 ((and (member test-status '("PREQ_FAIL" "PREQ_DISCARDED" "BLOCKED" "ZERO_ITEMS" "KEEP_TRYING" "TEN_STRIKES" "TIMED_OUT")))
                                  (rmt:set-state-status-and-roll-up-items run-id (db:test-get-id test) 'foo "NOT_STARTED" "n/a" (conc "kill-run moved from "test-state":"test-status" to NOT_STARTED:n/a"))
                                  ;;(mt:test-set-state-status-by-id run-id (db:test-get-id test) "NOT_STARTED" "n/a" (conc "kill-run moved from "test-state":"test-status" to NOT_STARTED:n/a"))
                                  (if (not (null? tal))
				      (loop (car tal)(cdr tal)))
                                  )
                                 (else
                                  (if (not (null? tal))
				      (loop (car tal)(cdr tal)))
                                  )))
			       ((set-state-status)
                                (let* ((new-state (car state-status))
                                       (new-status (cadr state-status))
                                       (test-id (db:test-get-id test))
                                       (test-run-dir (db:test-get-rundir new-test-dat))
                                       (has-subrun (and (subrun:subrun-test-initialized? test-run-dir)
                                                      (not (subrun:subrun-removed? test-run-dir)))))
                                  (when has-subrun
                                     (common:send-thunk-to-background-thread
                                        (lambda ()
                                          (subrun:set-state-status test-run-dir state status new-state-status)
                                          )
                                        )
                                    )
                                  (debug:print-info 2 *default-log-port* "new state " new-state ", new status " new-status )
                                  (mt:test-set-state-status-by-id run-id test-id new-state new-status #f))
				(if (not (null? tal))
				    (loop (car tal)(cdr tal))))
			       ((run-wait)
                                ;; BB TODO - manage has-subrun case
				(debug:print-info 2 *default-log-port* "still waiting, " (length tests) " tests still running")
				(thread-sleep! 5)
				(let ((new-tests (proc-get-tests run-id)))
				  (if (null? new-tests)
				      (debug:print-info 1 *default-log-port* "Run completed according to zero tests matching provided criteria.")
				      (loop (car new-tests)(cdr new-tests)))))
			       ((archive)
                                ;; BB TODO - manage has-subrun case
				(if (and run-dir (not toplevel-with-children))
				    (let ((ddir (conc run-dir "/")))
				      (case (string->symbol (args:get-arg "-archive"))
					((save save-remove keep-html)
					 (if (common:file-exists? ddir)
					     (debug:print-info 0 *default-log-port* "Estimating disk space usage for " test-fulln ": " (common:get-disk-space-used ddir)))))))
				(if (not (null? tal))
				    (loop (car tal)(cdr tal))))
			       )))
		       )
		     (if worker-thread (thread-join! worker-thread)))
                   (common:join-backgrounded-threads))))
	   
	   ;; remove the run if zero tests remain
	   (if (eq? action 'remove-runs)
	       (let* ((run-id   (db:get-value-by-header run header "id")) ;; NB// masks run-id from above?
                      (remtests (mt:get-tests-for-run run-id #f '("DELETED") '("n/a") not-in: #t)))
		 (if (null? remtests) ;; no more tests remaining
	           (let* ((linkspath (remove-last-path-directory lasttpath))
                          (runpaths (hash-table-keys run-paths-hash))
                    )

                    (debug:print 2 *default-log-port* "run-paths-hash: " (hash-table-keys run-paths-hash))

                    (debug:print 1 *default-log-port* "Removing target " target "run: " run-name)
                    (if (not keep-records)
                      (begin
                        (debug:print 1 *default-log-port* "Removing DB records for the run.")
                        (rmt:delete-run run-id)
                        (rmt:delete-old-deleted-test-records))
                    )
                    (if (not (equal?  linkspath "/does/not/exist/I"))
	               (begin 
                         (debug:print 1 *default-log-port* "Recursively removing links dir " linkspath)
                         (runs:recursive-delete-with-error-msg linkspath)))

                   (for-each (lambda(runpath)
                       (debug:print 1 *default-log-port* "Recursively removing runs dir " runpath)
                       (runs:recursive-delete-with-error-msg runpath)
                     )
                     runpaths
                   )
		       )))))
	 ))
     runs)
    ;; special case - archive get
    (if (equal? (args:get-arg "-archive") "get")
	(archive:bup-get-data "get" #f #f test-records rp-mutex bup-mutex))
    (if (or (equal? (args:get-arg "-archive") "save") (equal? (args:get-arg "-archive") "save-remove"))
        (begin 
             (print "db archive started")  
             (archive:megatest-db target runnamepatt)
             (print "db archived")))
    )
  #t
  )

;;======================================================================
;; Routines for manipulating runs
;;======================================================================

;; Since many calls to a run require pretty much the same setup 
;; this wrapper is used to reduce the replication of code
(define (general-run-call switchname action-desc proc)
  (let ((runname (common:args-get-runname))
	(target  (common:args-get-target)))
    (cond
     ((not target)
      (debug:print-error 0 *default-log-port* "Missing required parameter for " switchname ", you must specify the target with -target")
      (exit 3))
     ((not runname)
      (debug:print-error 0 *default-log-port* "Missing required parameter for " switchname ", you must specify the run name with -runname runname")
      (exit 3))
     (else
      (let (;; (db   #f)
	    (keys #f))
	(if (launch:setup)
	    (begin
	      (full-runconfigs-read) ;; cache the run config
	      ;; (launch:cache-config) ;; there are two independent config cache locations, turning this one off for now. MRW.
	      ) ;; do not cache here - need to be sure runconfigs is processed
	    (begin 
	      (debug:print 0 *default-log-port* "Failed to setup, exiting")
	      (exit 1)))

        
	(set! keys (keys:config-get-fields *configdat*))
	;; have enough to process -target or -reqtarg here
	(if (args:get-arg "-reqtarg")
	    (let* ((runconfigf (conc  *toppath* "/runconfigs.config")) ;; DO NOT EVALUATE ALL 
		   (runconfig  (configf:read-config runconfigf #f #t environ-patt: #f env-to-use: (module-environment 'bigmod))))
	      (if (hash-table-ref/default runconfig (args:get-arg "-reqtarg") #f)
		  (keys:target-set-args keys (args:get-arg "-reqtarg") args:arg-hash)
		    
		  (begin
		    (debug:print-error 0 *default-log-port* "[" (args:get-arg "-reqtarg") "] not found in " runconfigf)
		    ;; (if db (sqlite3:finalize! db))
		    (exit 1)
		    )))
	    (if (args:get-arg "-target")
		(keys:target-set-args keys (args:get-arg "-target" args:arg-hash) args:arg-hash)))
	(if (not (car *configinfo*))
	    (begin
	      (debug:print-error 0 *default-log-port* "Attempted to " action-desc " but run area config file not found")
	      (exit 1))
	    ;; Extract out stuff needed in most or many calls
	    ;; here then call proc
	    (let* ((keyvals    (keys:target->keyval keys target)))
	      (proc target runname keys keyvals)))
	;; (if db (sqlite3:finalize! db))
	(set! *didsomething* #t))))))

;;======================================================================
;; Lock/unlock runs
;;======================================================================

(define (runs:handle-locking target keys runname lock unlock user)
  (let* ((db       #f)
	 (rundat   (mt:get-runs-by-patt keys runname target))
	 (header   (vector-ref rundat 0))
	 (runs     (vector-ref rundat 1)))
    (for-each (lambda (run)
		(let ((run-id (db:get-value-by-header run header "id"))
           (str (if lock 
									"lock"
                  "unlock")))
		  (if (or lock
			  (and unlock
			       (or (args:get-arg "-force")
                (begin
				 (print "Do you really wish to unlock run " run-id "?\n   y/n: ")
				 (equal? "y" (read-line))))))
          (begin 
		      (rmt:lock/unlock-run run-id lock unlock user)
					(debug:print-info 0 *default-log-port* "Done " str " on run id " run-id))
		      (debug:print-info 0 *default-log-port* "Skipping lock/unlock on " run-id))))
	      runs)))
;;======================================================================
;; Rollup runs
;;======================================================================

;; Update the test_meta table for this test
(define (runs:update-test_meta test-name test-conf)
  (let ((currrecord (rmt:testmeta-get-record test-name)))
    (if (not currrecord)
	(begin
	  (set! currrecord (make-vector 11 #f))
	  (rmt:testmeta-add-record test-name)))
    (for-each 
     (lambda (key)
       (let* ((idx (cadr key))
	      (fld (car  key))
	      (val (configf:lookup test-conf "test_meta" fld)))
	 ;; (debug:print 5 *default-log-port* "idx: " idx " fld: " fld " val: " val)
	 (if (and val (not (equal? (vector-ref currrecord idx) val)))
	     (begin
	       (print "Updating " test-name " " fld " to " val)
	       (rmt:testmeta-update-field test-name fld val)))))
     '(("author" 2)("owner" 3)("description" 4)("reviewed" 5)("tags" 9)("jobgroup" 10)))))

;; find tests with matching tags, tagpatt is a string "tagpatt1,tagpatt2%, ..."
;;
(define (runs:get-tests-matching-tags tagpatt)
  (let* ((tagdata (rmt:get-tests-tags))
         (res     '())) ;; list of tests that match one or more tags
    (for-each
     (lambda (row)
       (let* ((tag (car row))
              (tests (cdr row)))
       (if (patt-list-match tag tagpatt)
           (set! res (append tests res)))))
     tagdata)
    res))
    

;; Update test_meta for all tests
(define (runs:update-all-test_meta db)
  (let ((test-names (tests:get-all))) ;; (tests:get-valid-tests)))
    (for-each 
     (lambda (test-name)
       (let* ((test-conf    (mt:lazy-read-test-config test-name)))
	 (if test-conf (runs:update-test_meta test-name test-conf))))
     (hash-table-keys test-names))))

;; This could probably be refactored into one complex query ...
;; NOT PORTED - DO NOT USE YET
;;
#;(define (runs:rollup-run keys runname user keyvals)
  (debug:print 4 *default-log-port* "runs:rollup-run, keys: " keys " -runname " runname " user: " user)
  (let* ((db              #f)
	 ;; register run operates on the main db
	 (new-run-id      (rmt:register-run keyvals runname "new" "n/a" user (args:get-arg "-contour")))
	 (prev-tests      (rmt:get-matching-previous-test-run-records new-run-id "%" "%"))
	 (curr-tests      (mt:get-tests-for-run new-run-id "%/%" '() '()))
	 (curr-tests-hash (make-hash-table)))
    (rmt:update-run-event_time new-run-id)
    ;; index the already saved tests by testname and itemdat in curr-tests-hash
    (for-each
     (lambda (testdat)
       (let* ((testname  (db:test-get-testname testdat))
	      (item-path (db:test-get-item-path testdat))
	      (full-name (conc testname "/" item-path)))
	 (hash-table-set! curr-tests-hash full-name testdat)))
     curr-tests)
    ;; NOPE: Non-optimal approach. Try this instead.
    ;;   1. tests are received in a list, most recent first
    ;;   2. replace the rollup test with the new *always*
    (for-each 
     (lambda (testdat)
       (let* ((testname  (db:test-get-testname testdat))
	      (item-path (db:test-get-item-path testdat))
	      (full-name (conc testname "/" item-path))
	      (prev-test-dat (hash-table-ref/default curr-tests-hash full-name #f))
	      (test-steps    (rmt:get-steps-for-test (db:test-get-id testdat)))
	      (new-test-record #f))
	 ;; replace these with insert ... select
	 (apply sqlite3:execute 
		db 
		(conc "INSERT OR REPLACE INTO tests (run_id,testname,state,status,event_time,host,cpuload,diskfree,uname,rundir,item_path,run_duration,final_logf,comment) "
		      "VALUES (?,?,?,?,?,?,?,?,?,?,?,?,?,?);")
		new-run-id (cddr (vector->list testdat)))
	 (set! new-testdat (car (mt:get-tests-for-run new-run-id (conc testname "/" item-path) '() '())))
	 (hash-table-set! curr-tests-hash full-name new-testdat) ;; this could be confusing, which record should go into the lookup table?
	 ;; Now duplicate the test steps
	 (debug:print 4 *default-log-port* "Copying records in test_steps from test_id=" (db:test-get-id testdat) " to " (db:test-get-id new-testdat))
	 (cdb:remote-run ;; to be replaced, note: this routine is not used currently
	  (lambda ()
	    (sqlite3:execute 
	     db 
	     (conc "INSERT OR REPLACE INTO test_steps (test_id,stepname,state,status,event_time,comment) "
		   "SELECT " (db:test-get-id new-testdat) ",stepname,state,status,event_time,comment FROM test_steps WHERE test_id=?;")
	     (db:test-get-id testdat))
	    ;; Now duplicate the test data
	    (debug:print 4 *default-log-port* "Copying records in test_data from test_id=" (db:test-get-id testdat) " to " (db:test-get-id new-testdat))
	    (sqlite3:execute 
	     db 
	     (conc "INSERT OR REPLACE INTO test_data (test_id,category,variable,value,expected,tol,units,comment) "
		   "SELECT " (db:test-get-id new-testdat) ",category,variable,value,expected,tol,units,comment FROM test_data WHERE test_id=?;")
	     (db:test-get-id testdat))))
	 ))
     prev-tests)))

;; clean cache files
(define (runs:clean-cache target runname toppath)
  (if target
      (if runname
	  (let* ((linktree (common:get-linktree)) ;; (if toppath (configf:lookup *configdat* "setup" "linktree")))
		 (runtop   (conc linktree "/" target "/" runname))
		 (files    (if (common:file-exists? runtop)
			       (append (glob (conc runtop "/.megatest*"))
				       (glob (conc runtop "/.runconfig*")))
			       '())))
	    (if (null? files)
		(debug:print-info 0 *default-log-port* "No cached megatest or runconfigs files found. None removed.")
		(begin
		  (debug:print-info 0 *default-log-port* "Removing cached files:\n    " (string-intersperse files "\n    "))
		  (for-each 
		   (lambda (f)
		     (handle-exceptions
			 exn
			 (debug:print 0 *default-log-port* "WARNING: Failed to remove file " f ", exn=" exn)
		       (delete-file f)))
		   files))))
	  (debug:print-error 0 *default-log-port* "-clean-cache requires -runname."))
      (debug:print-error 0 *default-log-port* "-clean-cache requires -target or -reqtarg")))

;; kill any runner processes (i.e. processes handling -runtests) that match target/runname
;; 
;; do a remote call to get the task queue info but do the killing as self here.
;;
(define (tasks:kill-runner target run-name testpatt)
  (let ((records    (rmt:tasks-find-task-queue-records target run-name testpatt "running" "run-tests"))
	(hostpid-rx (regexp "\\s+(\\w+)\\s+(\\d+)$"))) ;; host pid is at end of param string
    (if (null? records)
	(debug:print 0 *default-log-port* "No run launching processes found for " target " / " run-name " with testpatt " (or testpatt "* no testpatt specified! *"))
	(debug:print 0 *default-log-port* "Found " (length records) " run(s) to kill."))
    (for-each 
     (lambda (record)
       (let* ((param-key (list-ref record 8))
	      (match-dat (string-search hostpid-rx param-key)))
	 (if match-dat
	     (let ((hostname  (cadr match-dat))
		   (pid       (string->number (caddr match-dat))))
	       (debug:print 0 *default-log-port* "Sending SIGINT to process " pid " on host " hostname)
	       (if (equal? (get-host-name) hostname)
		   (if (process:alive? pid)
		       (begin
			 (handle-exceptions
			  exn
			  (begin
			    (debug:print 0 *default-log-port* "Kill of process " pid " on host " hostname " failed.")
			    (debug:print 0 *default-log-port* " message: " ((condition-property-accessor 'exn 'message) exn))
			    #t)
			  (process-signal pid signal/int)
			  (thread-sleep! 5)
			  (if (process:alive? pid)
			      (process-signal pid signal/kill)))))
		   ;;  (call-with-environment-variables
		   (let ((old-targethost (get-environment-variable "TARGETHOST")))
		     (set-environment-variable! "TARGETHOST" hostname)
		     (set-environment-variable! "TARGETHOST_LOGF" "server-kills.log")
		     (system (conc "nbfake kill " pid))
		     (if old-targethost (set-environment-variable! "TARGETHOST" old-targethost))
		     (unset-environment-variable! "TARGETHOST")
		     (unset-environment-variable! "TARGETHOST_LOGF"))))
	     (debug:print-error 0 *default-log-port* "no record or improper record for " target "/" run-name " in tasks_queue in main.db"))))
     records)))

(define (task:get-run-times)
   (let* ( 
           (run-patt (if (args:get-arg "-run-patt")
                        (args:get-arg "-run-patt")
                        "%"))
           (target-patt (if (args:get-arg "-target-patt")
                        (args:get-arg "-target-patt")
                        "%"))
 
           (run-times  (rmt:get-run-times  run-patt target-patt )))
   (if (eq? (length run-times) 0)
     (begin
       (print "Data not found!!")
       (exit)))
   (if (equal? (args:get-arg "-dumpmode") "json")
       (task:print-runtime-as-json run-times)
         (if (equal? (args:get-arg "-dumpmode") "csv")
	     (task:print-runtime run-times ",")
	     (task:print-runtime run-times "  ")))))



)