#general
- w
witty-crayon-22786
03/08/2016, 5:46 PMyes, you would need to do that. but not very hard. - b
bored-art-40741
03/08/2016, 5:46 PMHow are these things hoisted into Resources() targets as it is now? - b
bored-art-40741
03/08/2016, 5:47 PMI'm confused about why it's difficult to differentiate between explicitly constructed user
and whatever is being magicked into the classpath by annotation processorsresources() - b
bored-art-40741
03/08/2016, 5:48 PMAnd really I mean magicked onto the graph - b
bored-art-40741
03/08/2016, 5:48 PMSince this is all about graph invalidation - a
ancient-stone-50795
03/08/2016, 5:48 PMThey aren't being magiced. Explicitly constructed user resources can be used as input to an annotation processor, and that's exactly how we use them. For example, a .xml file may contain information used to generate jaxb bindings. - b
bored-art-40741
03/08/2016, 5:48 PMI see - a
ancient-stone-50795
03/08/2016, 5:49 PMWe have an explicit plugin to generate code for that, but you could also write that as an annotation processor - b
bored-art-40741
03/08/2016, 5:49 PMHow difficult would it be for those to become explicitly constructed
, a trivial subclass?compile_resources - w
witty-crayon-22786
03/08/2016, 5:49 PMthat only partially solves the problem - w
witty-crayon-22786
03/08/2016, 5:49 PMthe annotation processor itself should not be on the runtime classpath - w
witty-crayon-22786
03/08/2016, 5:49 PMlet alone the resources it consumes - w
witty-crayon-22786
03/08/2016, 5:49 PMso the entire subgraph would be compile-time only - c
curved-manchester-66006
03/08/2016, 5:50 PMhmm thanks for looking - b
bored-art-40741
03/08/2016, 5:50 PMHow does the runtime classpath come into what I'm talking about here? - b
bored-art-40741
03/08/2016, 5:50 PMThe major issue I have with this is compile time invalidation - w
witty-crayon-22786
03/08/2016, 5:51 PMthis is about more than resources... it's also about other targets - w
witty-crayon-22786
03/08/2016, 5:52 PMsee the linked blog post, in particular its mention of pants. - b
bored-art-40741
03/08/2016, 5:52 PMtbh the overarching design issues here are less important to me right now. This is an invalidation bug that wastes a ton of user time for us. I'm happy to find some sort of upstream compromise solution, but right now I'm also just as happy to fork and eliminate this behavior entirely until someone upstream figures it out - w
witty-crayon-22786
03/08/2016, 5:53 PMyou should commit the fix you had on that ticket. it's orthogonal to this discussion, iirc - b
bored-art-40741
03/08/2016, 5:53 PMI believe that fix breaks Eric - w
witty-crayon-22786
03/08/2016, 5:53 PMno more than the current situation. - b
bored-art-40741
03/08/2016, 5:53 PMReally? cc @ancient-stone-50795 re the question above and also ^^ - b
bored-art-40741
03/08/2016, 5:54 PMThis RB I think is what you're talking about @witty-crayon-22786 https://rbcommons.com/s/twitter/r/3185/diff/1#index_header - b
bored-art-40741
03/08/2016, 5:55 PMOh wait, no - b
bored-art-40741
03/08/2016, 5:55 PMThat was different - w
witty-crayon-22786
03/08/2016, 5:55 PMthat is the review i'm referring to. - b
bored-art-40741
03/08/2016, 5:55 PMerrr yeah, that's the review. - w
witty-crayon-22786
03/08/2016, 5:55 PMthat is a bugfix, and does not affect the behaviour in master - w
witty-crayon-22786
03/08/2016, 5:55 PMchanging the behaviour in master is another thing