#general
- e
enough-analyst-54434
08/14/2015, 4:56 PMThis would only make sense if you rolled the cert recently or upgraded from 80 to 443 - a
ancient-stone-50795
08/14/2015, 4:56 PMI’m backtracking on that. Those directories were old dirs that have apparently moved. Data is being written into the cache. - e
enough-analyst-54434
08/14/2015, 4:57 PMk - a
ancient-stone-50795
08/14/2015, 4:57 PMRegardless, failing to HEAD a cache file shouldn’t cause pants to fail. - a
ancient-stone-50795
08/14/2015, 4:58 PMI have been working in plaintext_reporter.py and I know that it ignores output from workunits that aren’t from the main thread, so that’s why were’ not seeing it. - a
ancient-stone-50795
08/14/2015, 5:00 PMHere’s the point where the error is being raised:Copy coderaise NonfatalArtifactCacheError('Failed to {0} {1}. Error: {2} {3}'.format(method, self._url_string(path), response.status_code, response.reason)) - w
witty-crayon-22786
08/14/2015, 5:24 PMso, we hit plenty of buildcache errors, and don't see failures because of them. that causing the background thread to die sounds like a recently introduced error - a
ancient-stone-50795
08/14/2015, 5:30 PMI’m filing an issue, I’ve got a stacktrace - a
ancient-stone-50795
08/14/2015, 5:30 PM - a
ancient-stone-50795
08/14/2015, 5:32 PMin artifact_cache.py on line 162, I think that NonfatalArtifactCacheError is escaping
and being handledcall_insert() - a
ancient-stone-50795
08/14/2015, 5:33 PMSo our
gets away from the artifact cache subsystem and gets handled through some generic handler.NonfatalArtifactCacheError - a
ancient-stone-50795
08/14/2015, 5:35 PMThere are a few outside callers of call_insert and they probably need to be using a wrapped version of that call. - a
ancient-stone-50795
08/14/2015, 5:44 PMI’ve got a fix - a
ancient-stone-50795
08/14/2015, 5:53 PM - e
enough-analyst-54434
08/14/2015, 7:42 PM@happy-kitchen-89482: hoping to get this one in the release, but it does tread on options: https://rbcommons.com/s/twitter/r/2622/ - h
happy-kitchen-89482
08/14/2015, 7:46 PMthat seems like not a small change - h
happy-kitchen-89482
08/14/2015, 7:46 PMI haven't gotten to it yet - e
enough-analyst-54434
08/14/2015, 7:47 PMAck - the options part is small, but the new PluginResolver + test ... ~80 of the delta - is large - h
happy-kitchen-89482
08/14/2015, 7:48 PMI'll try and get to it now - e
enough-analyst-54434
08/14/2015, 7:48 PMThanks - I appreciate it - w
witty-crayon-22786
08/14/2015, 7:50 PMwill wait to see what happens there. we'd appreciate it for our next internal release. - e
enough-analyst-54434
08/14/2015, 7:54 PMWould pants_internal use that to bootstrap a pex? ... I'm not seeing how that would work - w
witty-crayon-22786
08/14/2015, 7:56 PMwe'd build a oss pants pex (or eggs, tbd), and then load the contrib modules we need via that mechanism - e
enough-analyst-54434
08/14/2015, 7:56 PMok - cool - w
witty-crayon-22786
08/14/2015, 7:56 PMand pants_internal would live in source as a pants_plugin target - e
enough-analyst-54434
08/14/2015, 7:56 PMNice - e
enough-analyst-54434
08/14/2015, 7:56 PMok - great - more users the better - w
witty-crayon-22786
08/14/2015, 7:57 PMprimary risk i can see is that if python resolution breaks, it breaks the contrib modules - w
witty-crayon-22786
08/14/2015, 7:57 PMer - w
witty-crayon-22786
08/14/2015, 7:57 PMyea, contrib modules or pants_plugin, i guess