Deploy planned for tomorrow: queues, expected result, starexec_build

We are planning a deploy for tomorrow which will fix a number of issues:

— we have improved code for interacting with external processes, which will fix our currently broken ability to manipulate our queues.  So I will be able to create queues for the upcoming competitions (with a test node in each), and move a few nodes which are not needed in QBF Gallery back into all.q.

— the bug with starexec-expected-result not being correctly taken into account when displaying solver results will be fixed.  This will retroactively fix problems with job pairs in previously executed jobs which have a starexec-result attribute, as long as the benchmark does indeed (currently, possibly after re-processing) have a starexec-expected-result attribute.

— community leaders (only) will have the ability to upload a solver archive containing a starexec_build script, which we will execute sandboxed to compile the solver (in other words, to generate whatever needs to be generated, which should then be invoked from a starexec_run_ script as usual).

— there are several other small bug fixes and usability improvements, too.

 

Posted in Uncategorized

Leave a comment