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

    enough-analyst-54434

    10/09/2015, 8:30 PM
    But multimodule pom, makes no sense to support easily imo
  • w

    witty-crayon-22786

    10/09/2015, 8:30 PM
    the next thing we need to be prepared for is migration gradle projects to pants 😃
  • w

    witty-crayon-22786

    10/09/2015, 8:30 PM
    lots o those around =P
  • h

    happy-kitchen-89482

    10/09/2015, 8:30 PM
    It's just a special case of build graph validation rules, and I see no need to single it out in this way
  • w

    witty-crayon-22786

    10/09/2015, 8:30 PM
    yep
  • e

    enough-analyst-54434

    10/09/2015, 8:30 PM
    I'm on board with that Benjy
  • h

    happy-kitchen-89482

    10/09/2015, 8:30 PM
    great
  • h

    happy-kitchen-89482

    10/09/2015, 8:30 PM
    then I shall cook something up
  • h

    happy-kitchen-89482

    10/09/2015, 8:31 PM
    my goal is that running './pants' should be instantaneous, and not, as now, spend a second parsing the bootstrap BUILD files before exiting.
  • w

    witty-crayon-22786

    10/09/2015, 8:41 PM
    @enough-analyst-54434: updated https://rbcommons.com/s/twitter/r/2937/ with the changes so far. will work on a more expedient fix.
  • w

    witty-crayon-22786

    10/09/2015, 8:42 PM
    ...that doesn't involve propagating Address all the way through
  • e

    enough-analyst-54434

    10/09/2015, 8:44 PM
    sg
  • e

    enough-analyst-54434

    10/09/2015, 8:44 PM
    I'm going to get a pex upgrade int for Nick's fix, just got it released
  • w

    witty-crayon-22786

    10/09/2015, 8:48 PM
    ok. will wait for that
  • w

    witty-crayon-22786

    10/09/2015, 8:57 PM
    @happy-kitchen-89482: one effect of the move of Address into build_graph is that the options package is going to depend on the build graph... which feels funky.
  • w

    witty-crayon-22786

    10/09/2015, 8:58 PM
    i think we agreed yesterday that the option types are the real culprit
  • w

    witty-crayon-22786

    10/09/2015, 8:58 PM
    but
  • e

    enough-analyst-54434

    10/09/2015, 8:58 PM
    But can't those option helpers move to where they are needed?
  • h

    happy-kitchen-89482

    10/09/2015, 8:58 PM
    I'm not sure options should have Address as a type, directly.
  • h

    happy-kitchen-89482

    10/09/2015, 8:58 PM
    I feel like it should deal with basic types only
  • w

    witty-crayon-22786

    10/09/2015, 8:58 PM
    hm, i disagree... it's also what sirois suggested on https://rbcommons.com/s/twitter/r/2937/
  • h

    happy-kitchen-89482

    10/09/2015, 8:59 PM
    I mean, that weird dep was always weird, we just hid the weirdness by putting Address in base and giving it its own target.
  • e

    enough-analyst-54434

    10/09/2015, 8:59 PM
    To be clear - option inputs basic yes, but immediately convert to internal model
  • w

    witty-crayon-22786

    10/09/2015, 8:59 PM
    there is a difference between a "valid parsed object" and an object that has links into the rest of the system
  • e

    enough-analyst-54434

    10/09/2015, 8:59 PM
    The dep is just totally artificial today
  • e

    enough-analyst-54434

    10/09/2015, 8:59 PM
    The special option converter here does not need to live in the file it lives in
  • w

    witty-crayon-22786

    10/09/2015, 8:59 PM
    right
  • h

    happy-kitchen-89482

    10/09/2015, 9:00 PM
    convert to internal model outside the options package though?
  • w

    witty-crayon-22786

    10/09/2015, 9:00 PM
    but to explode it out we'd need a "real interface"
  • e

    enough-analyst-54434

    10/09/2015, 9:00 PM
    yes
1...533534535...2420Latest