https://pantsbuild.org/ logo
Powered by
#development
  • w

    witty-crayon-22786

    06/14/2018, 1:42 AM
    ok
  • a

    aloof-angle-91616

    06/14/2018, 1:42 AM
    the only difference is that we set the platform and add a resource (the shared library)
  • w

    witty-crayon-22786

    06/14/2018, 1:42 AM
    ok
  • a

    aloof-angle-91616

    06/14/2018, 1:43 AM
    i'm trying to think of more
  • w

    witty-crayon-22786

    06/14/2018, 1:43 AM
    ok.
  • a

    aloof-angle-91616

    06/14/2018, 1:43 AM
    if it passes travis, i'm not any more worried about this affecting other python code than i am about anything else i do ever
  • w

    witty-crayon-22786

    06/14/2018, 1:44 AM
    heh.
  • w

    witty-crayon-22786

    06/14/2018, 1:44 AM
    twitter's 3rdparty can throw interesting wrenches into python changes, in general. but i think that you're right that this should not affect that.
  • a

    aloof-angle-91616

    06/14/2018, 1:46 AM
    there's a single failing test on #5815 which is just an export integration test, the last few comments on there describe it. i think it has something to do with my extension of the 
    DependencyContext
    you expanded when zinc 1.0 was merged. i was going to look at it once i fixed the not-glibc-anymore pr
  • w

    witty-crayon-22786

    06/14/2018, 1:46 AM
    @aloof-angle-91616: pinged on internal slack
  • a

    aloof-angle-91616

    06/14/2018, 3:28 AM
    it looks like the most minimal way to get the not-glibc-anymore pr's goals achieved are to do a search for the glibc libraries on the host -- this is i believe absolutely the right direction (thanks for the discussion in this channel!)
  • a

    aloof-angle-91616

    06/14/2018, 3:56 AM
    @witty-crayon-22786 do you have any thoughts on what might be causing this error invoking zinc on the export integration test? i made changes to the jvm 
    DependencyContext
    to use a base class, but I don't know at all what might cause zinc to fail to parse args: https://github.com/pantsbuild/pants/pull/5815#issuecomment-396424884
  • r

    rough-minister-58256

    06/14/2018, 3:56 AM
    👍
  • r

    rough-minister-58256

    06/14/2018, 3:57 AM
    that would definitely be more consistent with what a user coming from a working model of c compilation on al inux box would expect today - those bins would only run on likewise hosts and be linked for that host type etc
  • a

    aloof-angle-91616

    06/14/2018, 3:57 AM
    actually, this is something that i can potentially spend some time weeding out of #5815 as well, potentially with far less effort, and do later
  • a

    aloof-angle-91616

    06/14/2018, 4:06 AM
    nvm i thought i could extract it out even if i didn't know what was happening but that is proving difficult too
  • a

    aloof-angle-91616

    06/14/2018, 4:31 AM
    lol "fixed" it except i still have no clue what changed
  • a

    aloof-angle-91616

    06/14/2018, 7:07 AM
    extremely annoyed because i can’t type any more tonight, but i have some code to find and fingerprint the host libc and i’m going to break it out into a subsystem base class so XCodeCLITools can consume the find/fingerprint logic too (basically analogous), and it’s not a ton of code. i don’t know why i’m seeing the timeout on that intellij export test in #5815 and will investigate tomorrow
  • a

    aloof-angle-91616

    06/14/2018, 7:08 AM
    (libc part is in a separarate pr)
  • r

    rough-minister-58256

    06/15/2018, 12:02 AM
    @rough-minister-58256 uploaded a file:

    engine UI prototype▾

  • w

    witty-crayon-22786

    06/15/2018, 12:03 AM
    heh, neat. is that just displaying random rule names?
  • r

    rough-minister-58256

    06/15/2018, 12:04 AM
    yeah, entirely synthesized rn
  • r

    rough-minister-58256

    06/15/2018, 12:04 AM
    and synchronous atm
  • r

    rough-minister-58256

    06/15/2018, 12:04 AM
    (but pure rust!)
  • w

    witty-crayon-22786

    06/15/2018, 12:05 AM
    oh, nice!
  • w

    witty-crayon-22786

    06/15/2018, 12:07 AM
    i spent a bunch of time profiling this week, and... async profiling is a pain in the ass
  • w

    witty-crayon-22786

    06/15/2018, 12:07 AM
    this might be neat though:
  • r

    rough-minister-58256

    06/15/2018, 12:07 AM
    what would be the best way to make this execute in parallel in terms of future integration w/ the engine? is it cool to have a one-off display thread driving something like this or should I try to make that play nice with futures to be able to run those on a tokio threadpool?
  • w

    witty-crayon-22786

    06/15/2018, 12:08 AM
    @witty-crayon-22786 uploaded a file: rule_graph.005.dot.pdf
  • w

    witty-crayon-22786

    06/15/2018, 12:09 AM
    @rough-minister-58256: so... all meaningful work now happens either on the tokio runtime's threadpool, or on a CpuPool in the case of IO
1...521522523...906Latest