https://pantsbuild.org/ logo
Powered by
#general
  • j

    jolly-chef-92794

    02/17/2016, 7:19 PM
    huh, okay
  • w

    witty-crayon-22786

    02/17/2016, 7:19 PM
    (...er, well: sources will always be published)
  • w

    witty-crayon-22786

    02/17/2016, 7:19 PM
    but javadoc/scaladoc are via the products
  • w

    witty-crayon-22786

    02/17/2016, 7:21 PM
    oh. even simpler solution.
  • w

    witty-crayon-22786

    02/17/2016, 7:21 PM
    have SimpleCodegenTask include the inputs in the output
  • j

    jolly-chef-92794

    02/17/2016, 7:22 PM
    hm
  • w

    witty-crayon-22786

    02/17/2016, 7:22 PM
    ...perhaps not generically.
  • w

    witty-crayon-22786

    02/17/2016, 7:22 PM
    i think 
    go
    would complain.
  • j

    jolly-chef-92794

    02/17/2016, 7:22 PM
    Yeah that would work actually. Maybe behind a flag
  • j

    jolly-chef-92794

    02/17/2016, 7:25 PM
    I think that would result in a synthetic 
    java_library()
    whose sources were both generated 
    .java
    files, and actual 
    .proto
    files.
  • j

    jolly-chef-92794

    02/17/2016, 7:25 PM
    Which might break? I am not sure.
  • j

    jolly-chef-92794

    02/17/2016, 7:26 PM
    I think I can get around this by using 
    resources()
    in this case though.
  • j

    jolly-chef-92794

    02/17/2016, 7:27 PM
    In that I can make a 
    resources()
    target reference all the raw proto files, and make the synthetic java library depend on that.
  • j

    jolly-chef-92794

    02/17/2016, 7:27 PM
  • w

    witty-crayon-22786

    02/17/2016, 7:28 PM
    well, that would get you a separate jar which would end up on your runtime classpath
  • w

    witty-crayon-22786

    02/17/2016, 7:28 PM
    which is probably not what you want
  • j

    jolly-chef-92794

    02/17/2016, 7:28 PM
    Oh, the resources aren’t published in the same jar as the classes?
  • w

    witty-crayon-22786

    02/17/2016, 7:28 PM
    ...er.
  • w

    witty-crayon-22786

    02/17/2016, 7:29 PM
    now i'm wondering.
  • w

    witty-crayon-22786

    02/17/2016, 7:29 PM
    ...hm. just realized this is different from the rest of the codebase.
  • w

    witty-crayon-22786

    02/17/2016, 7:29 PM
    yes: for publishing, the resources would end up in the same jar
  • w

    witty-crayon-22786

    02/17/2016, 7:29 PM
    everywhere else (
    ./pants bundle
    , etc) they would be in a separate jar
  • w

    witty-crayon-22786

    02/17/2016, 7:30 PM
    either way, they would be on your runtime classpath
  • w

    witty-crayon-22786

    02/17/2016, 7:30 PM
    rather than in a 
    -sources
    classified jar
  • e

    enough-analyst-54434

    02/17/2016, 9:55 PM
    Validation for all of our option strife: https://docs.docker.com/machine/get-started-cloud/ : 
    --amazonec2-access-key
    --digitalocean-access-token
  • e

    enough-analyst-54434

    02/17/2016, 9:55 PM
    amazonec2
    and 
    digitalocean
    being plugins
  • f

    fancy-queen-20734

    02/17/2016, 10:19 PM
    discovers bare tuple syntax
  • b

    bored-art-40741

    02/17/2016, 10:46 PM
    Validation?
  • b

    bored-art-40741

    02/17/2016, 10:47 PM
    Is the suggestion that we use this or that this whole process is messy and difficult to provision?
  • w

    witty-crayon-22786

    02/17/2016, 10:49 PM
    i interpreted that as: "the new options system kicks ass, and was worth the effort in order to avoid something like this"
1...985986987...2420Latest