[ruby-core:35976] Re: require performance on 1.9

On 3/05/11 8:37 AM, KOSAKI Motohiro wrote:
> Hello,
>
> 2011/5/1 Xavier Shay<[e-mail]>:
>> Hello,
>> I am looking into the performance issues requiring files in 1.9.
>> Here is a graph I am trying to fix:
>> https://skitch.com/xaviershay/rh64d/tc-load-time
>> https://gist.github.com/c8d0d422a9203e1fe492 (after creating 10000 files in
>> the `f` dir)
>>
>> A commit from a few years back [1] adds some behaviour around
>> loaded_feature_path. When disabled, the require time in my benchmark is
>> halved [2]. There is no test covering this behaviour, and I can't quite
>> figure out what it is doing.
>>
>> It is probably explained in ruby-dev:31932 but I can't read Japanese:
>> http://markmail.org/message/d3ef53c63kx2xqkr
>>
>> Can anyone either translate the above message for me, or provide a test case
>> that covers the expected behaviour?
>
> I don't understand the issue. then I'll make simple translate. :-)
>
> : As I wrote [ruby-dev:31923], If I wrote
> :
> :----<hoge.rb>-------------------------------
> :class HOGE; end
> :--------------------------------------------
> :
> :and, autoload :HOGE, 'hoge.rb' is set on ruby-1.9, require 'hoge.rb'
> :makes following exception.
> :
> :------------------------------------------------------------------------
> :NameError: uninitialized constant HOGE
> :        from /home/nagai/Ruby-work/hoge.rb:1:in `<top (required)>'
> :------------------------------------------------------------------------
> :
> :ruby-1.8 works completely.
> :I think it's 'autoload' vs 'require' confliction issue. If I use
> HOGE.new instead of
> :'require', I've seen no issue.
>
>
> Is this related for your issue?
I can't make a test case that fails when I continue at load.c:161, 
skipping an entire code path. I have tried the following, the behaviour 
was identical with and without my patch.

----

autoload :HOGE, 'hoge.rb'
require 'hoge.rb'

# no such file to load -- hoge.rb (LoadError)
# This is expected since . isn't in load path
# Adding . to load path via -I option results in no error

---

autoload :HOGE, 'hoge.rb'
require './hoge.rb'
# Same as above, but fails on line 1 of hoge.rb. I thought this may be 
the problem but fails on both patched and unpatched rubies.

-----

autoload :HOGE, './hoge.rb'
require './hoge.rb'
# Works

-----

In short, I have a patch that eliminates a major code path for requiring 
that speeds it up twofold, and can't find a failing case for it. See: 
https://gist.github.com/950154 (now with graph!)

I am running this on a moderately sized rails application and seeing a 
*seven* second improvement in start up time (17s down to 10s).

Next weekend I am going to revisit the require code and really try to 
nut out what it's doing.

Xavier

>
>
>>
>> Cheers,
>> Xavier
>>
>> P.S. This may or many not be related to this issue:
>> http://redmine.ruby-lang.org/issues/show/3924
>>
>> [1]
>> https://github.com/ruby/ruby/commit/bb1f1c57823758d2450d184ce7c85beb0d538bf0
>>
>> [2]  add continue; to load.c:161 gives 3s for n=2500
>>
>>
>