#development
- w
witty-crayon-22786
05/11/2016, 3:31 AMsubject equals tuple of two things because method not updated to expect a tuple return type - w
witty-crayon-22786
05/11/2016, 3:31 AM...and bundle not actually covered in a test =x - r
rough-minister-58256
05/11/2016, 3:31 AMheh - w
witty-crayon-22786
05/11/2016, 3:32 AMwas so excited to ship the one bit of code today that i did a shoddy job - r
rough-minister-58256
05/11/2016, 3:33 AMhaha - w
witty-crayon-22786
05/11/2016, 3:35 AM@rough-minister-58256: one thing potentially related to your "previous run of :: causes future runs to fail with same error": i don't see the `Scheduler`'s debug line about runtime in runs that contain errors - w
witty-crayon-22786
05/11/2016, 3:36 AMwhich doesn't make a lot of sense to me, because we fully exit the scheduler regardless of `Throw`/`Return`... so that line should be hit consistently, even for noops - w
witty-crayon-22786
05/11/2016, 3:37 AMperhaps because the raised exception there aborts the fork? - w
witty-crayon-22786
05/11/2016, 3:37 AMbut still, by then the scheduler has exited - w
witty-crayon-22786
05/11/2016, 3:37 AMand should have logged - w
witty-crayon-22786
05/11/2016, 3:38 AMso, to repro: introduce an error in a file, then run - r
rough-minister-58256
05/11/2016, 3:38 AM - w
witty-crayon-22786
05/11/2016, 3:38 AMon re-run, i would expect to see that line still... it would have run nothing... but still should log (EDIT) - w
witty-crayon-22786
05/11/2016, 3:39 AM@rough-minister-58256: yea, can see in your example: no "ran X scheduling iterations" line - r
rough-minister-58256
05/11/2016, 3:39 AMyeah usually it spits outran 0 scheduling iterations - r
rough-minister-58256
05/11/2016, 3:39 AMon successful runs - r
rough-minister-58256
05/11/2016, 3:39 AMwill take a look at that one - w
witty-crayon-22786
05/11/2016, 3:40 AMif we fail to fork, the graph we built is thrown away, right - w
witty-crayon-22786
05/11/2016, 3:40 AM? - r
rough-minister-58256
05/11/2016, 3:41 AMthe graph building happens well before forking - r
rough-minister-58256
05/11/2016, 3:41 AMso it either provides a graph or None - w
witty-crayon-22786
05/11/2016, 3:41 AMmm - r
rough-minister-58256
05/11/2016, 3:42 AM - r
rough-minister-58256
05/11/2016, 3:43 AM
starts as None, gets assigned hte output ofbuild_graph
so if that excepts its still None - as if scheduler wasn’t even thereSchedulerService.get_build_graph() - w
witty-crayon-22786
05/11/2016, 3:56 AM@rough-minister-58256: repro for the client hang i was mentioning: https://github.com/pantsbuild/pants/issues/3390 - r
rough-minister-58256
05/11/2016, 4:05 AMnice thx - b
billowy-beach-68498
05/11/2016, 4:52 AM@witty-crayon-22786: in your today’s commit, I have questions about file_map test, for example, exclude_globs test - b
billowy-beach-68498
05/11/2016, 4:53 AMit seems exlucde=globs(“*bb.py”) is treated as rglobs? - w
witty-crayon-22786
05/11/2016, 4:53 AM@billowy-beach-68498: the excludes were still suffering from the bug that eric had pointed out - w
witty-crayon-22786
05/11/2016, 4:53 AMhm, shouldn't be.