GitLab

Slow the example down a bit and allow 2 Workers. In "-can_specialize 1" mode the two Workers will complete the whole pipeline.

setting "-hive_force_init 1" will cause init_pipeline.pl to drop the database prior to creation (use with care!)

added support for stringification/destringification of accumulated values (an element is allowed to be a complex structure)

now supports sleeping for a floating point seconds; take_time can be given by a runtime-computed formula such as "1+rand(1)/1000"

Simplified logic to decide whether Scheduler needs a resync. Temporarily ignore limiters and look at the number of workers initially required.

Trigger re-sync when sum_over_preferred_analyses(num_required_workers)==0.
Trigger check-for-dead if valley->count_running_workers!=queen->count_running_workers.

finally implemented LSF's version of count_running_workers() and a Valley aggregator for all visible meadows

(currently they repeat a cycle of "wait t=1 sec, refresh stats" n=5 times,
maybe both of these parameters could be tuned better in future)

All Hive scripts now detect $::hive_root_dir and use it for setting the @INC so manual setting of PERL5LIB is only needed if using API directly

make sure beekeeper.pl runs runWorker.pl from its own scripts directory (ignore the one in the path)

Copied the @-tag annotation from tables.sql to tables.sqlite. Unlike the original mysql version, the SQLite version gives no warnings when processed by sql2html.pl

added a new protected_prepare_execute() method to avoid deadlocks and used it twice in AnalysisJobAdaptor, to fix Stephen's deadlocks

bugfix: do not crash on encountering accumulated dataflow (just ignore it for the moment); work correctly in DisplayStretched mode

bugfix: check before storing rc (may be necessary in -analysis_topup mode) and warn about consequences of redefining it.

API extension: store() now also returns how many actual store operations (as opposed to fetching of already stored ones) it has performed