Megatest

  hed tal reg reruns  test-record
  test-name item-path jobgroup
  waitons testmode  newtal itemmaps prereqs-not-met)

;; set up needed environment variables given a run-id and optionally a target, itempath etc.
;;
(define (runs:set-megatest-env-vars run-id #!key (inkeys #f)(inrunname #f)(inkeyvals #f)(intarget #f)(testname #f)(itempath #f))
  ;;(bb-check-path msg: "runs:set-megatest-env-vars entry")
  (let* ((target    (or intarget 
			(common:args-get-target)
			(get-environment-variable "MT_TARGET")))
	 (keys      (if inkeys    inkeys    (rmt:get-keys)))
	 (keyvals   (if inkeyvals inkeyvals (keys:target->keyval keys target)))
	 (vals      (hash-table-ref/default *env-vars-by-run-id* run-id #f))
	 (link-tree (common:get-linktree))) ;; (configf:lookup *configdat* "setup" "linktree")))
    (if testname (setenv "MT_TEST_NAME" testname))
    (if itempath (setenv "MT_ITEMPATH"  itempath))

    ;; get the info from the db and put it in the cache
    (if link-tree
	(setenv "MT_LINKTREE" link-tree)
	(debug:print-error 0 *default-log-port* "linktree not set, should be set in megatest.config in [setup] section."))
    (if (not vals)
	(let ((ht (make-hash-table)))
	  (hash-table-set! *env-vars-by-run-id* run-id ht)
	  (set! vals ht)
	  (for-each
	   (lambda (key)
	     (hash-table-set! vals (car key) (cadr key)))
	   keyvals)))
    ;; from the cached data set the vars
    
    (hash-table-for-each
     vals
     (lambda (key val)
       (debug:print 2 *default-log-port* "setenv " key " " val)
       (safe-setenv key val)))
    ;;(bb-check-path msg: "runs:set-megatest-env-vars block 1")
    ;;(BB> "*env-vars-by-run-id*/runid("run-id" vals="(hash-table->alist vals))

    (if (not (get-environment-variable "MT_TARGET"))(setenv "MT_TARGET" target))
    ;; we had a case where there was an exception generated by the hash-table-ref
    ;; due to *configdat* being #f Adding a handle and exit
    (let fatal-loop ((count 0)) 
      (handle-exceptions
	  exn
	  (let ((call-chain (get-call-chain))
		(msg        ((condition-property-accessor 'exn 'message) exn)))
	    (if (< count 5)
		(begin ;; this call is colliding, do some crude stuff to fix it.
		  (debug:print 0 *default-log-port* "ERROR: *configdat* was inaccessible! This should never happen. Retry #" count)
		  (launch:setup force-reread: #t)
		  (fatal-loop (+ count 1))) 
		(begin
		  (debug:print 0 *default-log-port* "FATAL: *configdat* was inaccessible! This should never happen. Retried " count " times. Message: " msg)
		  (debug:print 0 *default-log-port* "Call chain:")
		  (with-output-to-port *default-log-port*

                    (lambda ()
                      (print "*configdat* is >>"*configdat*"<<")
                      (pp *configdat*)
                      (pp call-chain)))
                  
		  (exit 1))))
          ;;(bb-check-path msg: "runs:set-megatest-env-vars block 1.5")
          (when (or (not *configdat*) (not (hash-table? *configdat*)))
              (debug:print 0 *default-log-port* "WARNING: *configdat* was inaccessible! This should never happen.  Brute force reread.")
              ;;(BB> "ERROR: *configdat* was inaccessible! This should never happen.  Brute force reread.")
              (thread-sleep! 2) ;; assuming nfs lag.
              (launch:setup force-reread: #t))
          (alist->env-vars (hash-table-ref/default *configdat* "env-override" '())))) ;;;; environment is tainted HERE in this let block.
    ;;(bb-check-path msg: "runs:set-megatest-env-vars block 2")
    ;; Lets use this as an opportunity to put MT_RUNNAME in the environment
    (let ((runname  (if inrunname inrunname (rmt:get-run-name-from-id run-id))))
      (if runname
	  (setenv "MT_RUNNAME" runname)
	  (debug:print-error 0 *default-log-port* "no value for runname for id " run-id)))
    (setenv "MT_RUN_AREA_HOME" *toppath*)
    ;; if a testname and itempath are available set the remaining appropriate variables
    (if testname (setenv "MT_TEST_NAME" testname))
    (if itempath (setenv "MT_ITEMPATH"  itempath))
    ;;(bb-check-path msg: "runs:set-megatest-env-vars block 3")
    (if (and testname link-tree)
	(setenv "MT_TEST_RUN_DIR" (conc (getenv "MT_LINKTREE")  "/"
					(getenv "MT_TARGET")    "/"
					(getenv "MT_RUNNAME")   "/"
					(getenv "MT_TEST_NAME")
					(if (and itempath
						 (not (equal? itempath "")))

  ;; 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))
  (thread-sleep! (cond
        	  ((> (runs:dat-can-run-more-tests-count runsdat) 20)
		   (if (runs:lownoise "waiting on tasks" 60)(debug:print-info 2 *default-log-port* "waiting for tasks to complete, sleeping briefly ..."))
                   (configf:lookup-number *configdat* "setup" "inter-test-delay" default: 0.1) ;; was 2
		   );; obviously haven't had any work to do for a while
        	  (else 0)))
  (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 (config-lookup *configdat* "jobgroups" jobgroup)))
				    (if (string? jobg-count)
					(string->number jobg-count)
					jobg-count))))

				  (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 (getenv "MT_TARGET") "/" "-") "-" (getenv "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)
                                         #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 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."))))))

;; 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)
  (let* ((tests-in-testpatt
          (map
           (lambda (test-patt-item)
             (car (string-split test-patt-item "/")))
           (string-split test-patt ",")))
         (waitors-upon-not-mentioned (lset-difference equal? waitors-upon tests-in-testpatt)))
    (null? waitors-upon-not-mentioned)))

;;  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* "/megatest.db"))
         (readonly-mode      (not (file-write-access? 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))

    ;; 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
    (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
		     (set! *time-to-exit* #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")

    ;; Now generate all the tests lists
    (set! all-tests-registry (tests:get-all))   ;; hash of testname => path-to-test

	  ;; 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)
    
    ;; 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))
	(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.
	  (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)) ;; waiton-tconfig below will be #f until that test is visted here at least once
		(hash-table-set! test-records
				 hed (vector hed     ;; 0
					     config  ;; 1
					     waitons ;; 2
					     (config-lookup config "requirements" "priority")     ;; priority 3
					     (tests:get-items config) ;; 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))) ;; BB: items expanded here - chained-waiton goes awry by now.
		     (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 ()				    
					    ;; (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

;;   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)

;; 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 2 *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

     ((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
		  (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* "runs:expand-items: (or (null? prereqs-not-met) (and (member 'toplevel testmode)(null? non-completed)))")
      (let ((test-name (tests:testqueue-get-testname test-record)))
	(setenv "MT_TEST_NAME" test-name) ;; 
	(setenv "MT_RUNNAME"   runname)

	(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)
	   (null? prereq-fails)
	   (null? non-completed))
      (if  (runs:can-keep-running? hed 20)
	  (begin
	    (runs:inc-cant-run-tests hed)
	    (debug:print-info 1 *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)
	    ;; 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 "COMPLETED" "PREQ_DISCARDED" "Failed to run due to prior failed prerequisites")
		(mt:test-set-state-status-by-id run-id test-id "COMPLETED" "PREQ_FAIL"      "Failed to run due to failed prerequisites"))))
      (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

	 (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
	 (numcpus                 (common:get-num-cpus #f))
	 (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") "1.2"))) ;; 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)

     ;; 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)))

      (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))
      ;; 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 maxload ;; only gate if maxload is specified
          (common:wait-for-cpuload maxload numcpus waitdelay))
      (if maxhomehostload
          (common:wait-for-homehost-load maxhomehostload (conc "Waiting for homehost load to drop below normalized value of " maxhomehostload)))
      
      (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

	  (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..."))
	    (thread-sleep! 1)
	    (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 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 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
		    (mt:test-set-state-status-by-testname 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)))
		    (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)
			  (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)
		      (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 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

  ;; 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

		      (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))

			   waitons:     waitons
			   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*))

	  (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))) ;; itemized test expanded here
		  (if loop-list
		      (apply loop loop-list)))
		;; 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* "Exiting loop with...\n  hed=" hed "\n  tal=" tal "\n  reruns=" reruns))
	 )))
    ;; 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")
		   (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))
		(begin
		  (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))
		  (rmt:find-and-mark-incomplete run-id #f)))
	    (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")) ;; TODO: pull from *common:stuff...*
		 (not (member (db:test-get-status test)
			      '("PASS" "WARN" "CHECK" "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-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))))

		(begin
		  (debug:print-info 0 *default-log-port* "WARNING: server is overloaded, trying again in one second")
		  (thread-sleep! 1)
		  (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

			(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"))))
		  ((and skip-check
			(configf:lookup test-conf "skip" "fileexists"))
		   (if (common:file-exists? (configf:lookup test-conf "skip" "fileexists"))
		       (set! skip-test (conc "Skipping due to existance of file " (configf:lookup test-conf "skip" "fileexists")))))

		  ((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))

;; 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 .")
        (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)

		    ;; 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-info 1 *default-log-port* "Run completed according to zero tests matching provided criteria.")
				      (loop (car new-tests)(cdr new-tests)))))
			       ((archive)
				(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))))))
	   ;; 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* ((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.
	(begin ;; let* ((realpath (resolve-pathname run-dir)))
	  (debug:print-info 1 *default-log-port* "Recursively removing " real-dir)
	  (if (common:file-exists? real-dir)
	      (runs:safe-delete-test-dir real-dir)
	      (debug:print 0 *default-log-port* "WARNING: test dir " real-dir " appears to not exist or is not readable")))
	(if real-dir 
	    (debug:print 0 *default-log-port* "WARNING: directory " real-dir " does not exist")
	    (debug:print 0 *default-log-port* "WARNING: no real directory corrosponding to link " run-dir ", nothing done")))
    (if (symbolic-link? run-dir)
	(begin

		 (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

      (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  (read-config runconfigf #f #t environ-patt: #f)))
	      (if (hash-table-ref/default runconfig (args:get-arg "-reqtarg") #f)
		  (keys:target-set-args keys (args:get-arg "-reqtarg") args:arg-hash)

	     (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)
		       (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")))