#general
- b
bored-art-40741
07/23/2015, 2:34 AMAgreed - w
witty-crayon-22786
07/23/2015, 2:34 AMaye - b
bored-art-40741
07/23/2015, 2:35 AMThere are several libraries I'd love to do this with right now - e
enough-analyst-54434
07/23/2015, 2:36 AMIts probably "easy" enough to rig internal ci to external red/green, but that would not be too satisfying as an OSS contributor - e
enough-analyst-54434
07/23/2015, 2:36 AMI'd want to see, perf loss, obfuscated stacktrace with just OSS code un-obfuscated - e
enough-analyst-54434
07/23/2015, 2:37 AMbasically enough data that the company wasn't just saying - trust us - your patch doesn't work for us - e
enough-analyst-54434
07/23/2015, 2:37 AMI think thats the single hardest problem to solve - e
enough-analyst-54434
07/23/2015, 2:40 AMThat and commit messages. You're in a monorepo so you take advantage and edit a core OSS lib and a sensitive industry secret internal lib that deps on it and you forget and talk about The GoldenGoose in the commit message - e
enough-analyst-54434
07/23/2015, 2:41 AM.... now syncing the OSS subslice got manual message scrubby - e
enough-analyst-54434
07/23/2015, 2:44 AMI dealt with the latter for twitter/commons back when I did the syncs and its alot of work and a fundamental limit on sync latency even if you sync per-commit - b
bored-art-40741
07/23/2015, 2:45 AMUnrelated: we're starting to get the strong impression internally that since our recent upgrade (a big one, we were like 6-8 weeks behind master) that we've started getting significantly fewer cache hits - b
bored-art-40741
07/23/2015, 2:46 AMWe're concerned that new fingerprinting code could be inadvertently invalidating targets that shouldn't be - e
enough-analyst-54434
07/23/2015, 2:46 AMSortof related: the company telling me my change broke their perf in ways hard to communicate! - e
enough-analyst-54434
07/23/2015, 2:46 AMnot fun for either side - b
bored-art-40741
07/23/2015, 2:46 AMYeah, I've asked for a repro, but it's a tough sell - e
enough-analyst-54434
07/23/2015, 2:46 AMyeah - b
bored-art-40741
07/23/2015, 2:47 AMBut given the nature of recent changes it seems plausible - e
enough-analyst-54434
07/23/2015, 2:48 AMagreed - w
witty-crayon-22786
07/23/2015, 2:49 AM@bored-art-40741: are you guys using different options in ci and on laptops? - w
witty-crayon-22786
07/23/2015, 2:49 AM(we haven't deployed cody's changes yet internally, so can't confirm/deny) - b
bored-art-40741
07/23/2015, 2:49 AMThe only options that differ are the cache reading/writing - w
witty-crayon-22786
07/23/2015, 2:49 AMyea - b
bored-art-40741
07/23/2015, 2:50 AMAlso we're on the release behind the one you're referring to - b
bored-art-40741
07/23/2015, 2:50 AMSo that one (thankfully) isn't in play as the culprit - w
witty-crayon-22786
07/23/2015, 2:50 AMer, he has two large changes - b
bored-art-40741
07/23/2015, 2:50 AMOh - b
bored-art-40741
07/23/2015, 2:50 AMWe're on upstream 0.0.34 - w
witty-crayon-22786
07/23/2015, 2:50 AMyea, i think you're behind both then. - b
bored-art-40741
07/23/2015, 2:51 AMI'm also suspecting that this might be an issue with our cache server and not fingerprinting semantics - w
witty-crayon-22786
07/23/2015, 2:51 AMmm