#general
- f
full-kilobyte-93429
09/30/2015, 9:51 PMyea, it's supposed to have been fixed by https://rbcommons.com/s/twitter/r/2672/ - b
bored-art-40741
09/30/2015, 9:51 PM(1) Do we need to use absolute paths on the classpath? - f
full-kilobyte-93429
09/30/2015, 9:51 PMby i overlooked the fact it calls executor which catches OSError - b
bored-art-40741
09/30/2015, 9:52 PM(2) Could we make a general purpose symlink farm that specifically provides super short (relative) paths in like.pants.d/symfarm/e8735e8773.jar -> .pants.d/compile/zinc/isolated-jars/... - b
bored-art-40741
09/30/2015, 9:52 PMOne time use nailgun also seems reasonable - b
bored-art-40741
09/30/2015, 9:52 PMI think that might be the "proper" fix when classpaths are just really big - w
witty-crayon-22786
09/30/2015, 9:52 PMsymlink farm would work, yea. - f
full-kilobyte-93429
09/30/2015, 9:53 PM@bored-art-40741: https://github.com/pantsbuild/pants/issues/1988 also has some past discussions on this topic - w
witty-crayon-22786
09/30/2015, 9:53 PM@bored-art-40741: well, it doesn't fully solve the problem. - w
witty-crayon-22786
09/30/2015, 9:53 PMjust delays the explosion some more - l
limited-country-20626
09/30/2015, 9:53 PMthat is a really clever fix - b
bored-art-40741
09/30/2015, 9:53 PMOh, I didn't realize what your strategy was here - b
bored-art-40741
09/30/2015, 9:53 PMThat's great - b
bored-art-40741
09/30/2015, 9:53 PMDefinitely a better way to go - l
limited-country-20626
09/30/2015, 9:54 PM@full-kilobyte-93429: Will you put me on the RB? - f
full-kilobyte-93429
09/30/2015, 9:54 PMyea, absolutely will probably find some time today or tomorrow - l
limited-country-20626
09/30/2015, 9:54 PMThanks all - w
witty-crayon-22786
09/30/2015, 9:56 PM@full-kilobyte-93429: if possible, would be great to not use exceptions for this case, and just maybe make it either 1) configurable with a low default or 2) always on - b
bored-art-40741
09/30/2015, 9:57 PMI think always on would be better - w
witty-crayon-22786
09/30/2015, 9:58 PMyea - b
bored-art-40741
09/30/2015, 9:58 PMIf there's an issue with it, having that issue only manifest after a particular magic number is exceeded that isn't user visible will be very hard to debug - w
witty-crayon-22786
09/30/2015, 9:58 PMas long as logging isn't affected, i agree - w
witty-crayon-22786
09/30/2015, 9:58 PMand the jar sticks around to be cracked open - w
witty-crayon-22786
09/30/2015, 9:59 PM(this calls to mind the fact that python tests are implemented as the building of a pex) - f
full-kilobyte-93429
09/30/2015, 10:05 PMhmm, always on when debugging you would do the extra step to examine the manifest - f
full-kilobyte-93429
09/30/2015, 10:06 PMit's a behavior change from today will need to educate users - w
witty-crayon-22786
09/30/2015, 10:07 PM@full-kilobyte-93429: no, not necessarily - w
witty-crayon-22786
09/30/2015, 10:07 PM@full-kilobyte-93429: not if it is always logged. - f
full-kilobyte-93429
09/30/2015, 10:07 PMlog the original path but execute with another? - w
witty-crayon-22786
09/30/2015, 10:07 PMand if the jar is preserved, you can still execute it