#general
- w
witty-crayon-22786
09/25/2015, 5:53 PMhave had some excellent discussions with @most-cricket-96651 over the last few days, since he's been in SF - w
witty-crayon-22786
09/25/2015, 5:54 PMhe suggested (independently) that the idea from the last summit of getting a prototype that spits out an action graph (even just for visual inspection) would be an excellent first step - e
enough-analyst-54434
09/25/2015, 5:54 PMyup - w
witty-crayon-22786
09/25/2015, 5:58 PMand @salmon-summer-38098 has been refreshing her thoughts in the engine doc - w
witty-crayon-22786
09/25/2015, 5:58 PMand i woke up at 5 this morning thinking about cross compilation D= - s
salmon-summer-38098
09/25/2015, 5:59 PM😄 - e
enough-analyst-54434
09/25/2015, 6:01 PMThe enigine is pretty easy to reconstruct in ~4 hours with 1 mildly interesting test case, say java8lib -> (thrift,java6lib) which was the case we used in the last whiteboard go-round on this - e
enough-analyst-54434
09/25/2015, 6:01 PMThat will raise all the thornier issues - w
witty-crayon-22786
09/25/2015, 6:02 PMthe scala[_js]_library one is an interesting testcase as well i think - e
enough-analyst-54434
09/25/2015, 6:05 PMOK - release is done so I'm going to recreate that case for my afternoon work and see if I can't throw up a strawman experiment RB to demonstrate whatever flows - w
witty-crayon-22786
09/25/2015, 6:05 PMexcellent. - s
salmon-summer-38098
09/25/2015, 6:12 PM@enough-analyst-54434: sounds good! - h
happy-kitchen-89482
09/25/2015, 6:14 PMyep, I have wanted that engine simulator forever. once that works, the rest of the effort is just filling in the blanks, and is quite distributable among us. - h
happy-kitchen-89482
09/25/2015, 6:15 PMAlso, hi @most-cricket-96651! Wilkommen. - h
happy-kitchen-89482
09/25/2015, 6:16 PMBasically the "only" difference between that simulator and real pants is the task code that transforms products to other products, and we mostly already know how to do that part pretty well. - m
most-cricket-96651
09/25/2015, 6:38 PMThanks. - l
limited-country-20626
09/25/2015, 6:42 PMSo I have a build that appears to be memory limited and going very slow. - l
limited-country-20626
09/25/2015, 6:43 PMThe CPU is running at about 30% but memory is almost entirely consumed. Activity manager shows 14.4 gbs of memory to python and growing. Any debug tips @witty-crayon-22786? - w
witty-crayon-22786
09/25/2015, 6:49 PMto python? hm, no idea! - w
witty-crayon-22786
09/25/2015, 6:49 PMyou could try dtruss or your os' equivalent - b
bored-art-40741
09/25/2015, 6:54 PMI was wondering if @rough-minister-58256 's profiling trick also collected memory stats - w
witty-crayon-22786
09/25/2015, 6:59 PM(you definitely haveCopy code
set, right?)[compile.jvm-dep-check] skip: True - b
bored-art-40741
09/25/2015, 7:00 PMYes - b
bored-art-40741
09/25/2015, 7:00 PMIt's unusably slow without that - b
bored-art-40741
09/25/2015, 7:00 PMThis seems to be something that happens with very large graphs - w
witty-crayon-22786
09/25/2015, 7:12 PMjust about time to kill some products - w
witty-crayon-22786
09/25/2015, 7:12 PMlookin at you next,classes_by_target - b
bored-art-40741
09/25/2015, 7:13 PMI wouldn't be surprised if that's the culprit - w
witty-crayon-22786
09/25/2015, 7:30 PM@bright-dinner-81701: did you see the feedback on https://rbcommons.com/s/twitter/r/2883/ ? - b
bored-art-40741
09/25/2015, 7:35 PMSo profiling python memory is still as much of a PITA as ever