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

    aloof-angle-91616

    10/01/2018, 4:30 PM
    daniel mentioned my large number of open WIP PRs which is something i have actually been continuously re-evaluating and if the fix in https://github.com/pantsbuild/pants/pull/6552 works i can then actually do other things including burning through that backlog. i have a PR on the above nailgun/pailgun thing which is WIPpier than i'd like it to be but i do think it would be useful so i'm going to push a WIP and this week or absolutely this weekend i can reduce the number of open PRs, i was just pretty excited when i realized i was maybe maybe accidentally fixing hangs
  • a

    aloof-angle-91616

    10/01/2018, 4:31 PM
    the json encoding PR (https://github.com/pantsbuild/pants/pull/6475) needs to be done for sure, for example, i just literally realized there was nonzero work left on that right as i was pre-empted by work on #6552 last week so i'm very pumped to not be under that pressure anymore. if there are others (e.g. the default values for datatypes) that i should look into that anyone would specifically find useful or that specifically require some more thought that i haven't addressed yet, please me know, that would imply that people use the code i write which is still wild to me
  • a

    aloof-angle-91616

    10/01/2018, 5:45 PM
    https://github.com/pantsbuild/pants/pull/6579 got too pumped about contextmanagers and wrote a justification for my actions last night
  • a

    aloof-angle-91616

    10/01/2018, 5:46 PM
    https://github.com/pantsbuild/pants/pull/6552 is green so after taking a shower i will evaluate how to best use this week
  • a

    aloof-angle-91616

    10/01/2018, 5:46 PM
    there's bazelconf on my birthday so i need to have done something monumental by then
  • s

    salmon-summer-38098

    10/01/2018, 6:09 PM
    @aloof-angle-91616 - let me know when you are ready for me to review your split up changes
    a
    • 2
    • 10
  • f

    fancy-queen-20734

    10/03/2018, 10:54 PM
    Exception: Failed to execute process: Error from remote execution: Internal: ""
    That’s fun.
  • a

    aloof-angle-91616

    10/04/2018, 6:09 AM
    that's why i'm so hype about 6552 (well, actually about stu's PR before it)
  • a

    aloof-angle-91616

    10/04/2018, 6:11 AM
    Copy code
    20:27:38 timestamp: 2018-10-04T03:27:34.217185
20:27:38 args: ['/redacted/.cache/pants/pex/bin/pants.pex/1.10.0rc1-9dcca268/pants.pex', 'redacted-goal']
20:27:38 pid: 172468
20:27:38 Exception caught: (<type 'exceptions.ValueError'>)
20:27:38   File "/redacted/runpy.py", line 174, in _run_module_as_main
20:27:38     "__main__", fname, loader, pkg_name)
20:27:38   File "/redacted/runpy.py", line 72, in _run_code
20:27:38     exec code in run_globals
20:27:38   File "__main__.py", line 25, in <module>
20:27:38     bootstrap_pex(__entry_point__)
20:27:38   File ".bootstrap/_pex/pex_bootstrapper.py", line 170, in bootstrap_pex
20:27:38     pex.PEX(entry_point).execute()
20:27:38   File ".bootstrap/_pex/pex.py", line 387, in execute
20:27:38     sys.stderr.flush()
20:27:38 
20:27:38 Exception message: I/O operation on closed file
  • a

    aloof-angle-91616

    10/04/2018, 6:12 AM
    like this immediately turned a scary strange error into oh it's just because pex likes writing directly to stderr do not pass go do not collect $200
  • a

    aloof-angle-91616

    10/04/2018, 6:16 AM
    internally, we'd see 
    python_dist()
    compilation fail, but the errors were just encoding errors when pex printed directly to stderr (meaning we couldn't see the error at all, we saw the exception message the encoding error while printing) -- this was a very annoying problem that kris fixed immediately (by googling it, lol, also bc he's a genius) by setting 
    LC_ALL=C
    so gcc doesn't use smart quotes (((((lol))))), but like i feel in those two cases there's this """impedance mismatch""" (i've never used that word before and unclear if it applies) between pex as a cli tool and pex as a library and i'm talking specifically about writing directly to stderr -- the above error is one i caused, but that write error without even checking if the file is closed means i need to resort to a hack to view my real error
    • 1
    • 1
  • a

    aloof-angle-91616

    10/04/2018, 6:19 AM
    probably a #pex discussion
    • 1
    • 1
  • a

    aloof-angle-91616

    10/04/2018, 6:20 AM
    considering a script to process a snippet of internal CI output and redact all paths to post elsewhere and realized that would also greatly improve the readability of the snippet
  • a

    aloof-angle-91616

    10/04/2018, 6:23 AM
    i guess i mentioned the pex thing above because it was funny to me that the extra logging we added to get fatal errors from pantsd-runner processes worked perfectly but of course we still had zero visibility into the issue because of something else. i don't think writing to stderr is obviously incorrect behavior, and of course, the above error wasn't happening before my changes, so like pex isn't doing anything wrong here i just need to check some filehandles in pants and i'm just whining i think
  • a

    aloof-angle-91616

    10/04/2018, 6:27 AM
    i just want to send one more message to say that i now realize this isn't at all about pex as a library vs pex as a cli tool because this is failing in pex bootstrap code, so this was (as is seen in the argv) a cli invocation. so that message is extremely backwards. sorry
  • l

    limited-country-20626

    10/04/2018, 5:24 PM
    I know this is a light abuse of the channel topic, but can anyone have advice on how I can audit my compile workers?
  • l

    limited-country-20626

    10/04/2018, 5:24 PM
  • l

    limited-country-20626

    10/04/2018, 5:25 PM
    we are now on the new engine, and working towards enabling pants.d
  • l

    limited-country-20626

    10/04/2018, 5:26 PM
    but I am concerned that the workers are blocking somehow. It is possible and/or likely that I just don't know the compile backend well enough and they are fine
  • l

    limited-country-20626

    10/04/2018, 5:26 PM
    but I would love to verify that and write a unit test two to protect it
  • l

    limited-country-20626

    10/04/2018, 5:26 PM
    how do you all monitor that?
    a
    • 2
    • 4
  • w

    witty-crayon-22786

    10/04/2018, 5:27 PM
    we've never had reason to suspect an issue there, so haven't looked at it in a while
  • a

    average-vr-56795

    10/05/2018, 4:09 PM
    @witty-crayon-22786 Why does 
    graph_visualize
    operation on a 
    Session
    not an 
    ExecutionRequest
    ?
  • w

    witty-crayon-22786

    10/05/2018, 4:12 PM
    i think that that's because we support visualizing outside the context of a particular execution...?
  • w

    witty-crayon-22786

    10/05/2018, 4:13 PM
    but that can be changed if need be
  • w

    witty-crayon-22786

    10/05/2018, 4:13 PM
    the semantics of how visualization works can definitely be changed
  • w

    witty-crayon-22786

    10/05/2018, 4:13 PM
    @average-vr-56795: ^
  • a

    average-vr-56795

    10/05/2018, 4:16 PM
    Cool, thanks!
  • w

    witty-crayon-22786

    10/05/2018, 4:16 PM
    do you guys have a handle on that one? i don't really have a solution in my mind to the trace/visualize issue
  • a

    average-vr-56795

    10/05/2018, 4:20 PM
1...578579580...906Latest