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

    aloof-angle-91616

    12/14/2017, 12:06 AM
    i can paste the full command line but it is copy pasted from the previous output and i have checked that it is the same
  • w

    witty-crayon-22786

    12/14/2017, 12:06 AM
    yea. i expect someone else will run into it soon and have time to debug it
  • w

    witty-crayon-22786

    12/14/2017, 12:07 AM
    for now, can just skip commit hooks.
  • a

    aloof-angle-91616

    12/14/2017, 12:07 AM
    thanks, good plan
  • a

    aloof-angle-91616

    12/14/2017, 12:09 AM
    also, adding 
    --force
    to the command line makes it work too, not just in 
    check_rust_formatting.sh
    , just to be clear
  • w

    witty-crayon-22786

    12/14/2017, 6:41 PM
    @rough-minister-58256: how are you feeling?
  • r

    rough-minister-58256

    12/14/2017, 7:22 PM
    feeling ok. my strategy of ‘feeding the cold’ + extra rest seems to be helping.
  • w

    witty-crayon-22786

    12/14/2017, 7:23 PM
    nice.
  • w

    witty-crayon-22786

    12/14/2017, 9:16 PM
  • w

    witty-crayon-22786

    12/14/2017, 9:16 PM
    brb.
  • r

    rough-minister-58256

    12/14/2017, 9:19 PM
    sweet
  • w

    witty-crayon-22786

    12/14/2017, 10:31 PM
    @salmon-summer-38098: Fo you think you'll get the process stuff out today?
  • s

    salmon-summer-38098

    12/14/2017, 10:31 PM
    yeah working on it furiously!
  • s

    salmon-summer-38098

    12/14/2017, 10:32 PM
    fixing error in one of the tests - 
    test_failed_output_conversion_propagates_throw
  • s

    salmon-summer-38098

    12/14/2017, 10:32 PM
    others pass though
  • s

    salmon-summer-38098

    12/14/2017, 10:32 PM
    so I am optimistic
  • s

    salmon-summer-38098

    12/14/2017, 10:34 PM
    in any case, I will send out the diff by tonight so we can see the overall impact
  • s

    salmon-summer-38098

    12/14/2017, 10:34 PM
    because it touches a lot at this stage
  • w

    witty-crayon-22786

    12/14/2017, 10:50 PM
    Great, thanks!
  • w

    witty-crayon-22786

    12/14/2017, 10:51 PM
    Would be useful to look at even before you fix the last test probably (unless you're expecting to overhaul it)
  • s

    salmon-summer-38098

    12/14/2017, 10:51 PM
    yeah ^ in any case, I will send out the diff by tonight so we can see the overall impact
  • r

    rough-minister-58256

    12/14/2017, 11:22 PM
    https://github.com/pantsbuild/pants/pull/5201 should be good for a final review - have been unable to repro a protocol error by moving the lock higher up in the pailgun handling stack.
  • r

    rough-minister-58256

    12/14/2017, 11:23 PM
    note that i’ve also removed the event.wait() here, so post-this we’ll be racing watchman startup (but the initial clean-all on version bump will be significantly faster etc..)
  • r

    rough-minister-58256

    12/14/2017, 11:26 PM
    open to improvements if folks think we should be locking e.g. the second request in a similar fashion to avoid run vs edit races during the initial watchman startup window
  • w

    witty-crayon-22786

    12/14/2017, 11:30 PM
    looking.
  • r

    rough-minister-58256

    12/14/2017, 11:41 PM
    seems like for watchman startup mitigation, its a tradeoff of ‘eventual consistency’ vs ‘slow initial run’
  • w

    witty-crayon-22786

    12/14/2017, 11:42 PM
    hm
  • r

    rough-minister-58256

    12/14/2017, 11:42 PM
    the best solution I can think of is to block on watchman startup iff len(product_graph) > 0
  • r

    rough-minister-58256

    12/14/2017, 11:42 PM
    i.e. for subsequent runs, where state would be cached
  • r

    rough-minister-58256

    12/14/2017, 11:42 PM
    and thus could be stale
1...411412413...906Latest