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

    salmon-summer-38098

    03/14/2017, 12:26 AM
  • w

    witty-crayon-22786

    03/14/2017, 12:27 AM
    nice 😃
  • s

    salmon-summer-38098

    03/14/2017, 12:27 AM
    🙂
  • s

    salmon-summer-38098

    03/14/2017, 12:27 AM
    without the self change its a pretty tiny change then so far
  • w

    witty-crayon-22786

    03/14/2017, 12:28 AM
    @salmon-summer-38098 : i think i changed my mind about doing it over here... you could tackle it if you want?
  • s

    salmon-summer-38098

    03/14/2017, 12:28 AM
    works with me
  • w

    witty-crayon-22786

    03/14/2017, 12:28 AM
    or not, up to you
  • s

    salmon-summer-38098

    03/14/2017, 12:28 AM
    happy to 🙂
  • w

    witty-crayon-22786

    03/14/2017, 12:28 AM
    cool, sounds good.
  • w

    witty-crayon-22786

    03/14/2017, 12:29 AM
    start by changing it in 
    trait Node
    , and then go from there
  • s

    salmon-summer-38098

    03/14/2017, 12:29 AM
    do you recommend leaving Node creation in there or removing it altogether for the final patch?
  • s

    salmon-summer-38098

    03/14/2017, 12:29 AM
    yep, sounds good
  • w

    witty-crayon-22786

    03/14/2017, 12:30 AM
    @salmon-summer-38098: i'd leave it there for now.
  • s

    salmon-summer-38098

    03/14/2017, 12:31 AM
    kk, sounds good
  • w

    witty-crayon-22786

    03/14/2017, 12:31 AM
    BUT... i think you can actually remove the implementation of 
    Node
    for the inlined types maybe...?
  • w

    witty-crayon-22786

    03/14/2017, 12:31 AM
    ...maybe not.
  • w

    witty-crayon-22786

    03/14/2017, 12:32 AM
    ah... it depends whether they are occasionally called without inlining. which they might be.
  • w

    witty-crayon-22786

    03/14/2017, 12:32 AM
    so yea, forget about that
  • s

    salmon-summer-38098

    03/14/2017, 12:32 AM
    yeah I think so, if I just create a new method instead of run?
  • s

    salmon-summer-38098

    03/14/2017, 12:32 AM
    a pub method
  • s

    salmon-summer-38098

    03/14/2017, 12:32 AM
    ah yeah, ok
  • w

    witty-crayon-22786

    03/14/2017, 12:32 AM
    yea, don't worry about that. it should be small potatoes.
  • s

    salmon-summer-38098

    03/14/2017, 12:50 AM
  • w

    witty-crayon-22786

    03/14/2017, 12:51 AM
    @salmon-summer-38098 : nice! you should be able to remove a few calls like 
    let node = self.clone()
    in that diff
  • s

    salmon-summer-38098

    03/14/2017, 12:51 AM
    yep, sg
  • r

    rough-minister-58256

    03/14/2017, 1:02 AM
    so I definitely see a bit of disparity between top-based mem usage reporting and gperftools
  • r

    rough-minister-58256

    03/14/2017, 1:03 AM
    @User uploaded a file:

    Pasted image at 2017-03-13, 6:02 PMâ–¾

  • r

    rough-minister-58256

    03/14/2017, 1:03 AM
    vs 
    Dumping heap profile to /tmp/pants_profile/pants_4042.0015.heap (12839 MB allocated cumulatively, 1162 MB currently in use)
  • r

    rough-minister-58256

    03/14/2017, 1:03 AM
    but just a side observation
  • w

    witty-crayon-22786

    03/14/2017, 1:04 AM
    yea, more in top. you saw that with pprof though too, right?
1...267268269...906Latest