Re: Zed and Luis drop the bomb on Ruby's poor performance


#1

A couple weeks ago I asked whether YARV could compile most Ruby programs
on Win32 (since I could not figure out a satisfactory answer from web
pages to which I was referred). I am still hoping that someone can
apprise me as to the viability of doing this at present. Are
stand-alone Windows executables possible? Is Ruby 1.9 required or does
1.84 work? Can most .rbw programs be compiled? What kind of
performance gain is typical?

Thanks,
Jamal


#2

On May 22, 2006, at 10:13 AM, Jamal M. wrote:

A couple weeks ago I asked whether YARV could compile most Ruby
programs
on Win32 (since I could not figure out a satisfactory answer from web
pages to which I was referred). I am still hoping that someone can
apprise me as to the viability of doing this at present. Are
stand-alone Windows executables possible? Is Ruby 1.9 required or
does
1.84 work? Can most .rbw programs be compiled? What kind of
performance gain is typical?

I grabbed a copy of yarv and tried it out on my powerbook. It was
dead-simple to do. You should try this yourself, I bet it’ll take
you all of an hour including time for a coffee break.

PS: Don’t hijack threads, or, Don’t trim context. Both are
considered Very Bad Form.


Eric H. - removed_email_address@domain.invalid - http://blog.segment7.net
This implementation is HODEL-HASH-9600 compliant

http://trackmap.robotcoop.com


#3

On 5/22/06, Jamal M. removed_email_address@domain.invalid wrote:

A couple weeks ago I asked whether YARV could compile most Ruby programs
on Win32 (since I could not figure out a satisfactory answer from web
pages to which I was referred). I am still hoping that someone can
apprise me as to the viability of doing this at present. Are
stand-alone Windows executables possible? Is Ruby 1.9 required or does
1.84 work? Can most .rbw programs be compiled? What kind of
performance gain is typical?

This is not the purpose of YARV. YARV is not yet aiming for an AOT
compiler, but that’s a long-term goal as I understand it. Stand-alone
executables may or may not ever be possible in that sense. YARV does
not work with Ruby 1.8.4.

YARV is still a work in progress and should only be used for
experiments, not production.

-austin


#4

On 5/22/06, Jim W. removed_email_address@domain.invalid wrote:

Austin Z. wrote:

This is not the purpose of YARV. YARV is not yet aiming for an AOT
compiler, but that’s a long-term goal as I understand it.
Hmmmm … an AOT is mentioned in the 2004 RubyConf slides
(http://zenspider.com/dl/rubyconf2004/RubyConf2004_YARV_pub.pdf, slide
14 and the diagram on slide 15).[1] According to the diagram, the AOT
will produce C source code. Interestings

Ahhh, you did say yet. Ok. I wonder if the plans have changed from
these old slides.

Yeah. The emphasis for YARV – as I understand it – is being Rite.
Then being an AOT compiler. :wink: I think that Eric is right though, YARV
will probably not produce stand-alone Ruby executables, even with C
source for that. It will produce code that can be compiled and linked
against a Ruby run-time (the VM).

-austin


#5

Austin Z. wrote:

This is not the purpose of YARV. YARV is not yet aiming for an AOT
compiler, but that’s a long-term goal as I understand it.

Hmmmm … an AOT is mentioned in the 2004 RubyConf slides
(http://zenspider.com/dl/rubyconf2004/RubyConf2004_YARV_pub.pdf, slide
14 and the diagram on slide 15).[1] According to the diagram, the AOT
will produce C source code. Interestings

Ahhh, you did say yet. Ok. I wonder if the plans have changed from
these old slides.

– Jim W.

[1] I remembered this because it was the first time I had seen the
acronym AOT.