Performance regression with 1.1rc5? #1591

Closed
svenfuchs opened this Issue Dec 15, 2011 · 12 comments

Projects

None yet

7 participants

Contributor

Performance dropped from 1.851sec to 4.226sec for https://github.com/travis-ci/travis-core

https://gist.github.com/038f8f2103b656f38880

I've seen similar results with other travis-* repositories before.

If that's useful at all, this is Mac OS X 10.7.2, Ruby 1.9.2 (https://gist.github.com/2f4299b6a277878b972b), Gems: https://gist.github.com/046cf2f587bea5472971

I've noticed a rails boot time increase of 9s to 19s with 1.1rc5.

nick@nite:~$ uname -a
Linux nite 3.1.0-1-amd64 #1 SMP Mon Nov 14 08:02:25 UTC 2011 x86_64 GNU/Linux
nick@nite:~$ ruby -v
ruby 1.8.7 (2011-06-30 patchlevel 352) [x86_64-linux]
Owner
hone commented Dec 15, 2011

I think this is a performance regression from 1.1.pre.8 and 1.1.pre.9. Do you guys see performance changes b/t the rc's?

I only tried the difference between 1.1.rc.5 (which is what was install when I tried gem install bundler --pre) and 1.0.21 which is the current stable.

Contributor

@hone yep, you're right:

bundler-1.1.pre.8

bundle install 1.86s user 0.17s system 99% cpu 2.044 total

bundler-1.1.pre.9

bundle install 3.79s user 0.16s system 99% cpu 3.958 total

... repeated this a couple of times with similar results each.

Owner
hone commented Dec 16, 2011

I did some digging today with @tenderlove and it looks like it might be related to 9d69030 introducing a lot of calls to method_missing. I'll need to pair with @indirect and dig into this more. I'm not sure we're going to be able to speed this up for 1.1.0 :(

Owner

Butbut... The cache that that commit clears doesn't even exist in 1.0 :(

On Dec 15, 2011, at 6:26 PM, Terence Leereply@reply.github.com wrote:

I did some digging today with @tenderlove and it looks like it might be related to 9d69030 introducing a lot of calls to method_missing. I'll need to pair with @indirect and dig into this more. I'm not sure we're going to be able to speed this up for 1.1.0 :(


Reply to this email directly or view it on GitHub:
#1591 (comment)

I noticed that too (#1508)

Owner

Okay, so Santiago (mostly or completely) fixed this in 9e1540c. Please test rc6 and let us know how things look now. Thanks!

Contributor

rc6 is not released, right? I've cloned master and built the gem locally. I'm still getting something like this:

$ gem list bundler

*** LOCAL GEMS ***

bundler (1.1.rc.5)
travis-ci $ time bundle check
The Gemfile's dependencies are satisfied
bundle check  4.13s user 0.10s system 100% cpu 4.225 total
$ gem list bundler

*** LOCAL GEMS ***

bundler (1.1.pre.8)
$ time bundle check
The Gemfile's dependencies are satisfied
bundle check  1.49s user 0.11s system 99% cpu 1.601 total
Owner

Sorry for the confusion, I was looking at a CI failure and failed to release last night. RC6 is out now, please test against it. The master branch is not currently up to date, since we are finishing the 1.1 release out of the 1-1-stable branch.

Contributor
rohit commented Dec 4, 2012

Was this fixed? I'm guessing yes. If so can the issue be closed? :)

Contributor
radar commented Aug 8, 2013

@rohit I would think so :)

@radar radar closed this Aug 8, 2013
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment