From: Dave Thomas <Dave@...>
Date: 2001-12-20T01:33:51+09:00
Subject: [ruby-talk:28956] Re: A Ruby programmer walked into a bar ...

"Harry Ohlsen" <harryo@zip.com.au> writes:

> You may feel I'm being a little alarmist, but I think that if I were
> a Windows programmer and found that such a trivial multithreaded
> program seemed to fail completely in my working environment, I'd be
> a little worried.
> 
> The real problem I see with this is that it would seem that
> something that is intended to be a standard part of Ruby (ie,
> OS-independent multithreading) doesn't work as one would expect on
> at least one OS, and certainly is not consistent across OSes.

You're absolutely right. Threading does not work properly under
Windows.  Unfortunately, threading is also one of those complex areas
of the interpreter. It cuts across whole swathes of code, so unless
you're an expert on interpreter internals, you'll be battling mightily
to understand it all.

Chad Fowler and I have been looking at various Rubicon failures under
Windows, and Chad's been fixing some of the problems (or arranging for
them to be fixed), but in order to get issues like the threading bugs
fixed we're going to need help from either Matz or Guy.

Even though I don't personally use Windows, I've been pushing for
several months now to try to find a way to improve Ruby's support of
Windows. I still feel that this is a priority. The tests are there:
even getting Rubicon to run cleanly under Windows would be a big
win. As a community, we just need to put some resources behind getting
it done. That doesn't necessarily mean we need Matz or Guy full time
on Windows. There are plenty of things besides threading that can be
looked at, and most of those things are accessible to Windows
programmers. We just need to start looking at them.


Cheers


Dave