From: "Bernard Delmée" <bdelmee@...>
Date: 2001-10-03T05:32:14+09:00
Subject: [ruby-talk:21955] Re: Let's work on Windows support

"Benoit Cerrina" <benoit.cerrina@writeme.com> wrote:

> This is very surprising, are you sure you used the best possible
> optimization on all platforms?
> My own experience with other applications (vim in particular)
> tend to go the other way with VC builds first then mingw then cygwin.
> Benoit

Alright, here are some facts and figures to backup these surprising
claims. You'll find hereafter a link to a page with detailed timings
of my test script with different win32 builds of the ruby interpreter.
Actually I was mistaking yesterday, and you'll see that the order in
this case is the reverse of what you cite, thus with the VC binary
performing even slower than the mingw one. I agree that this is
completely counter-intuitive.

Caveat emptor: I had prepared this page about a month ago, mostly
because I was unahppy with the performance of my port of a perl script
to ruby. So I went on and compared it with two more languages. Later
on though, I discovered that performance almost was on par with python
when switching from the mingw-compiled ruby to the cygwin one.

In the end I never published the URL anywhere, since the last thing
I wanted was to ignite yet anoter silly computer language flame.
Especially against ruby !-}

Now let's just forget about the script itself and what it does, and
hope it can be recycled as a test case (amongst many others) for
comparing win32 ruby ports.

Read the whole story at http://www.bdelmee.easynet.be/lilshoot/

HTH,

Bernard.