Megatest

                   (debug:print 0 *default-log-port* "Message: " ((condition-property-accessor 'exn 'message) 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.")))))
    
(define (runs:run-post-hook run-id)
    (let* ((run-post-hook   (configf:lookup *configdat* "runs" "post-hook"))
           (existing-tests (if run-post-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 "post-hook-" (string-translate (getenv "MT_TARGET") "/" "-") "-" (getenv "MT_RUNNAME") ".log"))
           (full-log-fname  (conc log-dir "/" log-file)))
      (if run-post-hook
          ;; (if (null? existing-tests)
          ;;    (debug:print 0 *default-log-port* "Skipping post-hook call \"" run-post-hook "\" as there are existing tests for this run.")))))
	  (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)
					#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))
		  (debug:print 0 *default-log-port* "ERROR: failed to run post-hook " run-post-hook ", check the log " log-file))
	      (debug:print-info 0 *default-log-port* "running run-post-hook: \"" run-post-hook "\", log is " actual-logf)
	      (system (conc run-post-hook " >> " actual-logf " 2>&1"))
	      (debug:print-info 0 *default-log-port* "post-hook \"" run-post-hook "\" took " (- (current-seconds) start-time) " seconds to run."))))))

;;  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))

					       (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)))

    ;;======================================================================
    ;; 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
	  (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.
	  (setenv "MT_TEST_NAME" hed) ;; 
	  (let*-values (((waitons waitors config)(tests:get-waitons hed all-tests-registry)))
	    (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))
		(hash-table-set! test-records ;; BB: we are doing a manual make-tests:testqueue
				 hed (vector hed     ;; 0 ;; testname
					     config  ;; 1 
					     waitons ;; 2 
					     (config-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 ;; 
					     )))
	    (for-each 
	     (lambda (waiton)
	       (if (and waiton (not (member waiton test-names)))
		   (let* ((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)))   ;; BB:  items expanded here.
		     (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 

		     ;; (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))))
	      (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)
							       (rmt:find-and-mark-incomplete 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! th1)
	    (set! keep-going #f)
	    (thread-join! th2)

(define (runs:queue-next-reg tal reg n regfull)
  (if regfull
      (cdr reg)
      (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

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

;; BB: for future reference - suspect target vars are not expanded to env vars at this point (item expansion using [items]\nwhatever [system echo $TARGETVAR] doesnt work right whereas [system echo #{targetvar}] does.. Tal and Randy have tix on this.  on first pass, var not set, on second pass, ok.  
(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*

     ((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 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

	      (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" "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))
      (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)) ;; 
	    ;; 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)
	    ;; 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 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 run-id test-id "NOT_STARTED" "PREQ_DISCARDED" "Failed to run due to prior failed prerequisites")
                (begin
                  (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))
      (if (or (not (null? reg))(not (null? tal)))
	   (list (car newtal)(append (cdr newtal) reg) '() reruns)
	  #f)) 
     ((null? runnables) #f) ;; if we get here and non-completed is null then it is all over.
     (else
      (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

	       (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))

	     (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

	  (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)))

      
      (run:test run-id run-info keyvals runname test-record flags #f test-registry all-tests-registry)
      (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)


	  #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

	    ;; 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..."))
	    (thread-sleep! 1)
	    (list (car newtal)(cdr newtal) reg reruns))
	  ;; the waiton is FAIL so no point in trying to run hed ever again
	  (if (or (not (null? reg))(not (null? tal)))
	      (if (or (vector? hed)  (not (null? fails))) ;; BB: why do we need a vector?  in my case, fails is populated (prereq failed), reg is not nul, and we really want to drop this one
		  (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 run-id test-id "NOT_STARTED" "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! 4)

		      (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 10)))
		      (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."))
			    (mt:test-set-state-status-by-testname run-id test-name item-path "NOT_STARTED" "KEEP_TRYING" #f)
			    (hash-table-set! test-registry hed 0)

			    (runs:loop-values newtal reg reglen regfull))))


		     (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)))))
	      ;; can't drop this - maybe running? Just keep trying
	      (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)))))))))



;; 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

  (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           (config-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))

			   testmode:    testmode
			   newtal:      newtal
			   itemmaps:    itemmaps
			   ;; prereqs-not-met: prereqs-not-met
			   )))
	(runs:dat-regfull-set! runsdat regfull)

        ;; -- removed BB 17ww28 - no longer needed.
	;; every 15 minutes verify the server is there for this run
	;; (if (and (common:low-noise-print 240 "try start server"  run-id)
	;; 	 (not (or (and *runremote*
	;; 		       (remote-server-url *runremote*)
	;; 		       (server:ping (remote-server-url *runremote*)))
	;; 		  (server:check-if-running *toppath*))))
	;;     (server:kind-run *toppath*))
	
	(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 "")

	  (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)) ;; BB - target vars are env vars here? to allow expansion of [items]\nsomething [system echo $SOMETARGVAR], which is wonky
	  (let ((can-run-more    (runs:can-run-more-tests runsdat run-id jobgroup max-concurrent-jobs)))
	    (if (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)))
		  (if loop-list
		      (apply loop loop-list)))
		;; if can't run more just loop with next possible test

	 ((not (null? tal))
	  (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 0 *default-log-port* "Have leftovers!")
	  (loop (car reg)(cdr reg) '() reruns))
	 (else
	  (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
    
    ;; 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)
    (thread-sleep! 5) ;; 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")

	    (if (not (eq? num-running prev-num-running))
		(debug:print-info 0 *default-log-port* "run-wait specified, waiting on " num-running " tests in RUNNING, REMOTEHOSTSTART or LAUNCHED state at " (time->string (seconds->local-time (current-seconds)))))
	    (thread-sleep! 5)
	    ;; (wait-loop (rmt:get-count-tests-running-for-run-id run-id) num-running))))
	    (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!
    (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"))
		 (not (member (db:test-get-status test)

       (if (not (string-search (regexp "testdat.db") f))
	   (runs:recursive-delete-with-error-msg fullname)))
     (+ 1 x))
   0 real-dir)
  ;; then the entire directory
  (runs:recursive-delete-with-error-msg real-dir))

;; 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)))
	 (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") "")))
    (print "Actions: " actions)
    (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"))
		  (for-each
		   (lambda (action)
		     (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 %"))))
		       ((archive)
			(if remove (system (conc precmd " megatest -archive save-remove -target " target " -runname " (simple-run-runname run) " -testpatt %"))))))
		   actions))))
	    sorted)))
     ;; (print "Sorted: " (map simple-run-event_time sorted))
     ;; (print "Remove: " (map simple-run-event_time to-remove))))
     (hash-table-keys runs-ht))
    runs-ht))

;; (define (runs:remove-all-but-last-n-runs-per-target target-patts runpatt num-to-keep)
;;   (let ((data (runs:get-all-but-most-recent-n-per-target target-patts runpatt num-to-keep)))
;;     (for-each
;;      (lambda (target)
;;        (let ((runs-to-remove (hash-table-ref data target )))
;;          (for-each
;;           (lambda (run)
;;             (print "megatest -remove-runs -target " target " -runname " (simple-run-runname run) " -testpatt %"))
;;           runs-to-remove)))
;;      (hash-table-keys data))))

;; Remove runs
;; fields are passing in through 
;; action:
;;    'remove-runs
;;    'set-state-status
;;
;; NB// should pass in keys?
;;
(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".

    (let* ((write-access-actions '(remove-runs set-state-status archive run-wait))
           (dbfile             (conc  *toppath* "/megatest.db"))
           (readonly-mode      (not (file-write-access? dbfile))))
      (when (and readonly-mode
                 (member action write-access-actions))
        (debug:print-error 0 *default-log-port* "megatest.db is readonly.  Cannot proceed with action ["action"] in which write-access isrequired .")

                      (remtests (mt:get-tests-for-run run-id #f '("DELETED") '("n/a") not-in: #t)))
		 (if (null? remtests) ;; no more tests remaining
		     (let* ((dparts  (string-split lasttpath "/"))
			    (runpath (conc "/" (string-intersperse 
						(take dparts (- (length dparts) 1))
						"/"))))
		       (debug:print 1 *default-log-port* "Removing run: " runkey " " (db:get-value-by-header run header "runname") " and related record")
                       (if (not keep-records)
                           (begin
                             (rmt:delete-run run-id)
                             (rmt:delete-old-deleted-test-records)))
                           ;; (rmt:set-var "DELETED_TESTS" (current-seconds))
		       ;; need to figure out the path to the run dir and remove it if empty
		       ;;    (if (null? (glob (conc runpath "/*")))
		       ;;        (begin
		       ;; 	 (debug:print 1 *default-log-port* "Removing run dir " runpath)
		       ;; 	 (system (conc "rmdir -p " runpath))))
		       )))))
	 ))
     runs)
    ;; (sqlite3:finalize! (db:delay-if-busy tdbdat))
    )
  #t)

(define (runs:remove-test-directory test mode) ;; remove-data-only)
  (let* ((run-dir       (db:test-get-rundir test))    ;; run dir is from the link tree
	 (real-dir      (if (common:file-exists? run-dir)
			    ;; (resolve-pathname run-dir)
			    (common:nice-path run-dir)
			    #f))
         (clean-mode    (or mode 'remove-all))
         (test-id       (db:test-get-id test))
        ;; (lock-key      (conc "test-" test-id))
        ;; (got-lock      (let loop ((lock        (rmt:no-sync-get-lock lock-key))
	;; 			     (expire-time (+ (current-seconds) 30))) ;; give up on getting the lock and steal it after 15 seconds
	;; 		    (if (car lock)
	;; 			#t
	;; 			(if (> (current-seconds) expire-time)
	;; 			    (begin
	;; 			      (debug:print-info 0 *default-log-port* "Timed out waiting for a lock to clean test with id " test-id)
	;; 			      (rmt:no-sync-del! lock-key) ;; destroy the lock
	;; 			      (loop (rmt:no-sync-get-lock lock-key) expire-time)) ;; 
	;; 			    (begin
	;; 			      (thread-sleep! 1)
	 ;; 			      (loop (rmt:no-sync-get-lock lock-key) expire-time)))))))
	 )
    (case clean-mode
      ((remove-data-only)(mt:test-set-state-status-by-id (db:test-get-run_id test)(db:test-get-id test) "CLEANING" "LOCKED" #f))
      ((remove-all)      (mt:test-set-state-status-by-id (db:test-get-run_id test)(db:test-get-id test) "REMOVING" "LOCKED" #f))
      ((archive-remove)  (mt:test-set-state-status-by-id (db:test-get-run_id test)(db:test-get-id test) "ARCHIVE_REMOVING" #f #f)))
    (debug:print-info 1 *default-log-port* "Attempting to remove " (if real-dir (conc " dir " real-dir " and ") "") " link " run-dir)
    (if (and real-dir 
	     (> (string-length real-dir) 5)
	     (common:file-exists? real-dir)) ;; bad heuristic but should prevent /tmp /home etc.

		 (delete-directory run-dir)))
	    (if (and run-dir
		     (not (member run-dir (list "n/a" "/tmp/badname"))))
		(debug:print 0 *default-log-port* "WARNING: not removing " run-dir " as it either doesn't exist or is not a symlink")
		(debug:print 0 *default-log-port* "NOTE: the run dir for this test is undefined. Test may have already been deleted."))
	    ))
    ;; Only delete the records *after* removing the directory. If things fail we have a record 
    (case clean-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))))
    ;; (rmt:no-sync-del! lock-key)
    ))

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