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

    happy-kitchen-89482

    08/27/2018, 6:13 PM
    Yeah, cffi and grpc have been the banes
  • h

    happy-kitchen-89482

    08/27/2018, 6:18 PM
  • h

    happy-kitchen-89482

    08/27/2018, 6:18 PM
    Didn’t want to make this part of the release, but I’ve confirmed that the release change is green with this.
  • a

    average-vr-56795

    08/27/2018, 8:03 PM
    I spent some time hacking together fixes for the things blocking us moving to the pure rust grpc implementation... One of them is https://github.com/tower-rs/tower-grpc/pull/78 and the other is still WIP. That will clean up a lot of the horrors
    w
    • 2
    • 4
  • h

    happy-kitchen-89482

    08/28/2018, 6:19 AM
    Looks like pantsd doesn’t refresh the graph if requirements.txt changes, in a 
    python_requirements()
    situation. Is this known?
  • e

    enough-analyst-54434

    08/28/2018, 1:25 PM
    Not sure, but I found no open issue so filed https://github.com/pantsbuild/pants/issues/6404
  • w

    witty-crayon-22786

    08/28/2018, 2:43 PM
    good find, thanks!
  • w

    witty-crayon-22786

    08/28/2018, 2:44 PM
    an alternative implementation would be to give it a 
    default_sources_glob
    for that file?
  • e

    enough-analyst-54434

    08/28/2018, 2:48 PM
    As things stand the file is only convention so that wouldn't cover all cases w/o a deprecation cycle on allowed file names. And then, default_* wouldn't help - we'd need to actually totally lock down the source.
  • a

    average-vr-56795

    08/28/2018, 2:56 PM
    Is there a reason the arg is called 
    requirements_relpath
    not 
    source
    ?
  • e

    enough-analyst-54434

    08/28/2018, 3:14 PM
    Not that I'm aware of. I tend to prefer verbose explicit names though so I'd guess that was the motivation, but it was maybe 4 years ago.
  • h

    happy-kitchen-89482

    08/28/2018, 3:54 PM
    Thanks, was going to file it after creating a repro, in case it was just an artifact of our repo somehow.
  • h

    happy-kitchen-89482

    08/28/2018, 3:54 PM
    Oh, but you’ve proven it to be true from first principles, nice.
  • w

    witty-crayon-22786

    08/28/2018, 3:58 PM
    could certainly translate it to 
    source
    in the macro
  • w

    witty-crayon-22786

    08/28/2018, 4:01 PM
    ah, i was forgetting the relationship between the target type and the macro.
  • w

    witty-crayon-22786

    08/28/2018, 4:07 PM
    @average-vr-56795: do you want to re-review https://github.com/pantsbuild/pants/pull/5521 before i land it? unfortunately it was old enough that i had to squash it, so the history isn't useful
  • a

    average-vr-56795

    08/28/2018, 4:07 PM
    I have it open in a tab, was about to do the review 🙂
  • w

    witty-crayon-22786

    08/28/2018, 4:09 PM
    it has roughly the same shape. i was briefly planning to try and do a "composition of Resettables" type thing, but didn't see an obvious way to hide the resettable type (in two years, this is the first time i've run into object safety, blargh.)
  • a

    average-vr-56795

    08/28/2018, 4:10 PM
    I’d be interested in minimal fails-to-compile code example of what you were trying to do which wouldn’t work, just as a curiousity (definitely not blocking anything)
  • w

    witty-crayon-22786

    08/28/2018, 4:10 PM
    sure.
  • a

    average-vr-56795

    08/28/2018, 4:12 PM
    @enough-analyst-54434 On https://github.com/pantsbuild/pants/pull/6405 - I quite like that in the bazel model every file is implicitly also a target - not suggesting any changes, just food for thought 🙂
    e
    • 2
    • 6
  • w

    witty-crayon-22786

    08/28/2018, 4:19 PM
    @average-vr-56795: one other bit of color on https://github.com/pantsbuild/pants/pull/5521 that makes it less scary: RwLock panics on reentrancy
  • w

    witty-crayon-22786

    08/28/2018, 4:20 PM
    so it's very challenging to get the 
    with_shutdown
    order incorrect. you panic if you do
  • w

    witty-crayon-22786

    08/28/2018, 4:20 PM
    it's easy to forget to shut something down, but... yea.
  • a

    average-vr-56795

    08/28/2018, 4:21 PM
  • w

    witty-crayon-22786

    08/28/2018, 4:22 PM
    and actually, it's not quite accurate: what i said above is accurate
  • w

    witty-crayon-22786

    08/28/2018, 4:22 PM
    you don't deadlock: you panic
  • w

    witty-crayon-22786

    08/28/2018, 4:22 PM
    will update.
  • a

    average-vr-56795

    08/28/2018, 4:22 PM
    Thanks 🙂
  • a

    average-vr-56795

    08/28/2018, 4:22 PM
    Why are we switching from lazily to eagerly initialising?
    w
    • 2
    • 7
1...571572573...906Latest