#general
- w
witty-crayon-22786
08/05/2015, 9:56 PMlikely related to computing the closure of all targets - w
witty-crayon-22786
08/05/2015, 9:57 PM@bored-art-40741: has a general solution to this problem ever come up? - w
witty-crayon-22786
08/05/2015, 9:57 PMmemoizing in the build graph, or similar - b
bored-art-40741
08/05/2015, 9:58 PMSo far we haven't actually seen perf issues due to build graph operations specifically, only due to greedy BUILD file operations (globs and especially rglobs) and to some extent parsing the BUILD files - b
bored-art-40741
08/05/2015, 10:00 PMTuning it hasn't really come up, no. Back in the day I considered adding some memoization, but it will have to be done carefully. I also poked around and looked for some pre-optimized graph libs but nothing seemed like it was going to be a big enough win for us, considering that our graphs aren't enormous (millions of nodes) and we're not doing particularly sophisticated operations on them - b
bored-art-40741
08/05/2015, 10:00 PMAre you positive it's taking repeated closures that's the issue? - w
witty-crayon-22786
08/05/2015, 10:00 PMnope! still profiling - c
cool-thailand-74745
08/05/2015, 10:00 PMCopy codeException message: unsupported operand type(s) for -: 'FilesetWithSpec' and 'list' while executing BUILD file FilesystemBuildFile(/Users/wickman/clients/aurora/src/main/python/apache/aurora/tools/BUILD) Loading addresses from 'src/main/python/apache/aurora/tools' failed. - b
bored-art-40741
08/05/2015, 10:00 PMAh okay - c
cool-thailand-74745
08/05/2015, 10:00 PMwhat's the suggested alternative for rglobs arithmetic? - b
bored-art-40741
08/05/2015, 10:00 PMrglobs takes params now to support this - r
rough-minister-58256
08/05/2015, 10:00 PM@cool-thailand-74745: paste? - r
rough-minister-58256
08/05/2015, 10:00 PMya - c
cool-thailand-74745
08/05/2015, 10:00 PM"several people are typing" - w
witty-crayon-22786
08/05/2015, 10:00 PM@cool-thailand-74745:exclude=[globs(..)] - c
cool-thailand-74745
08/05/2015, 10:00 PMah - e
enough-analyst-54434
08/05/2015, 10:01 PMWe have really nice deprecation warnings spelling things like that out but obviously not helpful if the upgrade stride steps over those - c
cool-thailand-74745
08/05/2015, 10:02 PMah, yeah, this was from 0.0.36 to master - c
cool-thailand-74745
08/05/2015, 10:02 PMi mean, there is no semver so it's unclear what constitutes a major change - e
enough-analyst-54434
08/05/2015, 10:02 PMThey all are - yeah - you still need to read changelog until we 1.0.0 - c
cool-thailand-74745
08/05/2015, 10:03 PMokeydoke - c
cool-thailand-74745
08/05/2015, 10:04 PMno CHANGES.rst in root -- i'm guessing changelog is git log? - r
rough-minister-58256
08/05/2015, 10:04 PM./src/python/pants/CHANGELOG.rst - c
cool-thailand-74745
08/05/2015, 10:04 PMty! - e
enough-analyst-54434
08/05/2015, 10:05 PMAlso on pypi and in pantsbuild.github.io - c
cool-thailand-74745
08/05/2015, 10:06 PMnever learned to read, etc - c
cool-thailand-74745
08/05/2015, 10:09 PMnaïve question but where is the change mentioned in CHANGELOG.rst? searched for 'exclude' and 'deprecat' and didn't see it. from what ver to what ver were there deprecation warnings? - w
witty-crayon-22786
08/05/2015, 10:09 PMScreen Shot 2015-08-05 at 3.09.29 PM.png - c
cool-thailand-74745
08/05/2015, 10:10 PMis the idea that you do 0.x -> 0.x+1 -> ... -> 0.x.N to upgrade? - e
enough-analyst-54434
08/05/2015, 10:10 PMYou also want to run this against your pants.ini https://github.com/pantsbuild/pants/blob/master/migrations/options/src/python/BUILD