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

    fast-nail-55400

    06/20/2017, 5:46 PM
    going to see if it is reproducible on linux
  • w

    witty-crayon-22786

    06/20/2017, 5:46 PM
    that was going to be my first guess. immediately segfaulting like that is definitely not expected.
  • w

    witty-crayon-22786

    06/20/2017, 5:46 PM
    also, do you repro on smaller contexts?
  • f

    fast-nail-55400

    06/20/2017, 5:46 PM
    no
  • f

    fast-nail-55400

    06/20/2017, 5:47 PM
    I'm able to compile targets just fine
  • w

    witty-crayon-22786

    06/20/2017, 5:48 PM
    @fast-nail-55400 : had discussed with @limited-country-20626 , but you guys might consider waiting until https://github.com/pantsbuild/pants/issues/3941 has landed before switching to v2
  • w

    witty-crayon-22786

    06/20/2017, 5:48 PM
    large repos will see some slowdown for commands* that affect "the whole repo"
  • f

    fast-nail-55400

    06/20/2017, 5:49 PM
    we can do that. @limited-country-20626 and I were discussing switching back to the legacy engine for the time being
  • f

    fast-nail-55400

    06/20/2017, 5:49 PM
    of course, if I can collect data on the issue for you, I'd like to do that
  • w

    witty-crayon-22786

    06/20/2017, 5:50 PM
    that would be good, yea. particularly if you can phrase it in terms of built-in commands
  • w

    witty-crayon-22786

    06/20/2017, 5:50 PM
    v2 changed looks more like 
    ./pants --changed-parent=master compile
  • w

    witty-crayon-22786

    06/20/2017, 5:51 PM
    (no need for explicit `cmd`+changed tasks... any task can be used with changed.)
  • f

    fast-nail-55400

    06/20/2017, 5:51 PM
    cool
  • w

    witty-crayon-22786

    06/20/2017, 5:54 PM
    for the record: we're using v2 by default on developer laptops (OSX) and on CI boxes (centos7)
  • w

    witty-crayon-22786

    06/20/2017, 5:55 PM
    the only thing we're not using it for is the graph inspection tasks that CI runs to determine what to rebuild.
  • w

    witty-crayon-22786

    06/20/2017, 6:59 PM
    @happy-kitchen-89482 : to correct my answer to yesterday's question a bit: it is possible to disable incremental compile completely. but it does not disable zinc
  • w

    witty-crayon-22786

    06/20/2017, 6:59 PM
    --no-compile-zinc-incremental
  • w

    witty-crayon-22786

    06/20/2017, 7:00 PM
    basically just uses a clean 
    results_dir
    for each run.
  • h

    happy-kitchen-89482

    06/20/2017, 7:01 PM
    That would probably avoid the trouble we’re having
  • h

    happy-kitchen-89482

    06/20/2017, 7:01 PM
    We’re all java, and not all that much of it, so I don’t think incremental compilation is much of a win in practice
  • h

    happy-kitchen-89482

    06/20/2017, 7:02 PM
    [all java as in no scala, we have plenty of python and go code]
  • w

    witty-crayon-22786

    06/20/2017, 7:02 PM
    @happy-kitchen-89482 : also, relevant: https://twitter.com/retronym/status/876968660752977920
  • h

    happy-kitchen-89482

    06/20/2017, 7:03 PM
    Sweet
  • w

    witty-crayon-22786

    06/20/2017, 7:03 PM
    so about 25% overhead for zinc on a cold compile =/
  • h

    happy-kitchen-89482

    06/20/2017, 7:34 PM
    Someone remind me how to glob over multiple file suffixes in a sources= stanza?
  • h

    happy-kitchen-89482

    06/20/2017, 7:35 PM
    Can sources be a list of globs?
  • w

    witty-crayon-22786

    06/20/2017, 7:38 PM
    yes
  • w

    witty-crayon-22786

    06/20/2017, 7:38 PM
    sources=globs('*.py', '*.pyx')
  • h

    happy-kitchen-89482

    06/20/2017, 8:39 PM
    The intellij plugin doesn’t recognize this as valid cc @wide-energy-11069
  • w

    witty-crayon-22786

    06/20/2017, 8:40 PM
    huh
1...176117621763...2420Latest