#general
- b
billowy-alarm-80641
09/15/2015, 5:53 PMmaybe that stopped using it awhile ago - b
bored-art-40741
09/15/2015, 5:53 PMAha, figured it out! - b
bored-art-40741
09/15/2015, 5:53 PMThankfully it is not a pants bug - b
bored-art-40741
09/15/2015, 5:54 PMIt's just something kind of crazy we do internally - b
billowy-alarm-80641
09/15/2015, 5:54 PMoh, n/m - b
billowy-alarm-80641
09/15/2015, 5:54 PMi see it in the reveiw now - b
bored-art-40741
09/15/2015, 5:55 PMUnfortunately this was probably muddying our perf testing too - b
bored-art-40741
09/15/2015, 5:58 PMMaybe not as much as we thought - b
bored-art-40741
09/15/2015, 5:58 PMCertainly not the clean tests - f
full-kilobyte-93429
09/15/2015, 6:14 PM@ancient-stone-50795: for https://rbcommons.com/s/twitter/r/2815/ our rest interface is still wip, what's your interface like? - f
full-kilobyte-93429
09/15/2015, 6:15 PMhaving another script is ok but can be saved at this point - a
ancient-stone-50795
09/15/2015, 6:56 PM@full-kilobyte-93429: It is proprietary and subject to change, so I don’t want to rely on it. - a
ancient-stone-50795
09/15/2015, 6:57 PMCan we mediate it through a script of some kind? How often will we want to re-validate the service discovery? - c
cold-lizard-88792
09/15/2015, 7:18 PMQuestion: I believe it's supposed to be a good practice to list all direct dependencies in BUILD target deps section and not to rely on transitive dependencies. Is that true and is there a written statement about it somewhere? - w
witty-crayon-22786
09/15/2015, 7:19 PMit's "good hygiene" within maven communities, but not widely followed - c
cold-lizard-88792
09/15/2015, 7:19 PM(I've had people arguing otherwise) - w
witty-crayon-22786
09/15/2015, 7:21 PMi don't know about javac, but scalac requires the transitive deps to be present - b
bored-art-40741
09/15/2015, 7:25 PMpants expects direct dependencies - b
bored-art-40741
09/15/2015, 7:26 PMThe "pants philosophy", at least as we follow it, is to list the dependencies are they are implied by source - b
bored-art-40741
09/15/2015, 7:26 PMBased on the nature of your build tool, pants internally will gather up the transitive closure to present the full transitive dependencies to the compiler - e
enough-analyst-54434
09/15/2015, 7:27 PMThe only valid argument against lining up deps with imports is if the import is forced on you by a 3rdparty API. IE: If you do not use guava, but the API returns ImutableList, you should not select an ImmutableList guava version, the API should - e
enough-analyst-54434
09/15/2015, 7:27 PMI think that's beyond debate? - e
enough-analyst-54434
09/15/2015, 7:28 PMBut its this API enforce other-dep that muddies the simpler advice - always declare a dep per import, and its probably not talking through this 1 exception case, which is common enough in practice, that leads to debate about the coarsers always do x or y - w
witty-crayon-22786
09/15/2015, 7:30 PMyea. type inference muddies that too. - e
enough-analyst-54434
09/15/2015, 7:30 PMwell, scala is crazy-pants! - w
witty-crayon-22786
09/15/2015, 7:30 PM@enough-analyst-54434: java has type inference too =P - c
cold-lizard-88792
09/15/2015, 7:30 PMI personally understand the advice and the exception. is there a linkable recommendation like http://pantsbuild.github.io/build_files.html#the-111-rule ? - e
enough-analyst-54434
09/15/2015, 7:31 PMOK - what does type infernce in java land have to do with this then? Curious - not seeing it - w
witty-crayon-22786
09/15/2015, 7:31 PM@cold-lizard-88792: no, i don't think there is. but i think that's part of the reason we want to work on the build graph server - e
enough-analyst-54434
09/15/2015, 7:31 PMSergey - the best way to understand the advice is to see what happens when you don;t follow it