#general
- w
witty-crayon-22786
10/15/2015, 7:02 PM - w
witty-crayon-22786
10/15/2015, 7:03 PM(but i think in this case the factoring is correct.) - c
cuddly-architect-27013
10/15/2015, 7:03 PMi don’t think it’s causing problems, just as i’m trying to reorganize the coverage stuff i’m looking at what i’m leaving behind…that method itself is only fifteen lines, but called from other methods etc. that could potentially be dropped if the runner handled it - c
cuddly-architect-27013
10/15/2015, 7:03 PMok, fair enough…yeah i was checking out the runner to see how the args were being used - c
cuddly-architect-27013
10/15/2015, 7:08 PMis there a set of benchmarks used to test perf on things like that? - c
cuddly-architect-27013
10/15/2015, 7:09 PMi.e. to validate that scanning the entire junit space is too expensive? - w
witty-crayon-22786
10/15/2015, 7:21 PM@cuddly-architect-27013: no. but it definitely is, heh. - w
witty-crayon-22786
10/15/2015, 7:22 PMmacaw-swift has ~2000 source modules, and 900 3rdparty modules - w
witty-crayon-22786
10/15/2015, 7:22 PMit contains at least 1.5mm classes - w
witty-crayon-22786
10/15/2015, 7:22 PMif you can do the lookup on 15 of those, rather than 1.5mm (with random IO)... - c
cuddly-architect-27013
10/15/2015, 7:24 PMyeah, i def believe that op is slower, but if you’re running all tests in macaw-swift anyway…eh, i trust your judgement on this, will leave it alone - c
cuddly-architect-27013
10/15/2015, 7:24 PMsimple smile - w
witty-crayon-22786
10/15/2015, 7:24 PM@cuddly-architect-27013: it's not about "all tests" - w
witty-crayon-22786
10/15/2015, 7:24 PMit's about not inspecting all classes - w
witty-crayon-22786
10/15/2015, 7:25 PM1.5mm classfiles, of which only the classfiles "created" by the target are relevant - w
witty-crayon-22786
10/15/2015, 7:26 PMadditionally, if you happen to have 4 targets in a context, you want to blame the correct target when a particular class/test fails - h
happy-kitchen-89482
10/15/2015, 7:26 PMOK, on second thought, I'm still not comfortable relying on capturing the language out of the path. It prevents things like
and/src/java/main
./src/java/test - w
witty-crayon-22786
10/15/2015, 7:27 PM@happy-kitchen-89482: not if fixed patterns are allowed? - h
happy-kitchen-89482
10/15/2015, 7:27 PMAnd this would require something like--source-root-patterns - w
witty-crayon-22786
10/15/2015, 7:27 PM(longest wins) - c
cuddly-architect-27013
10/15/2015, 7:27 PM@witty-crayon-22786: ok, fair enough, thx for explaining - h
happy-kitchen-89482
10/15/2015, 7:27 PMyeah but if you have many of those, fixed patterns suck - h
happy-kitchen-89482
10/15/2015, 7:27 PMOr rather, fixed paths - h
happy-kitchen-89482
10/15/2015, 7:28 PM--source-root-patterns are the intermediate - w
witty-crayon-22786
10/15/2015, 7:28 PMk. still not sure it's worth having 5 options over - h
happy-kitchen-89482
10/15/2015, 7:28 PMyeah, I'm not super-keen on that - w
witty-crayon-22786
10/15/2015, 7:28 PMbut will ship! - h
happy-kitchen-89482
10/15/2015, 7:28 PMI would like to simplify - w
witty-crayon-22786
10/15/2015, 7:28 PMi think discouraging is fine here. "i need to define fixed roots when my roots are weird"... yea. - h
happy-kitchen-89482
10/15/2015, 7:28 PMbut right now I fear failing to capture important cases...