On Fri, 2003-05-30 at 21:43, Greg Fenton wrote:
> Agreed on the performance.  I have not had a "buggy" episode in quite
> some time.

Yes, stability does seem to have improved over the years.  But, I still
don't trust it.  ;-)  Performance is usually pretty bad.  But, it would
be fine for testing.  Performance might not be too terrible since I'm
using only native Win32 calls.

> Having SWISH-E behave nicely in Cygwin would complete my
> development/test cycle.

Sounds good to me.

> > I should clarify that you would also need native libxml2, pcre, and
> > zlib builds.
>
> I'm quite sure that all this compiles under Cygwin.

Yes, building them would be much simpler under Cygwin since you have a
full Unix filesystem.

> The big problem from what I can tell (though, I don't claim to be a
> Cygwin developer...just a user) is that the swish-e blindly converts
> command paths using '\' on _WIN32.  I think simply implementing Bill's
> suggestion (PATH_SEPARATOR) would work, but I have not looked into
> this.

Can we just disable make_windows_path() for Cygwin by modifying the
existing _WIN32 ifdef?  Have you found any other problems?

> So, should I consider working on a patch?

If you want.  I can try to incorporate it before 2.4.0 releases.




Bill, does Cygwin support for 2.4.0 sound like a good plan?  I bet it's
a simple change.

-- 
 David Norris
  http://www.webaugur.com/dave/
  ICQ - 412039