perl-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Randy Kobes <ra...@theoryx5.uwinnipeg.ca>
Subject Re: [mp2] source_scan on Win32
Date Fri, 18 Jun 2004 17:18:23 GMT
On Fri, 18 Jun 2004, Stas Bekman wrote:

> Randy Kobes wrote:
> > I've been trying to get the build/source_scan.pl working
> > on Win32, and made some progress, but came up against
> > something. First off, there's a couple of things that
> > are fixable:
> > - in lib/Apache/ParseSource.pm, follow => 1 is used
> > within a File::Find, which doesn't work on Win32;
> > - in lib/ModPerl/CScan.pm, a different syntax is
> > needed to run 'cppstdin':
[ ... ]
> Please don't create any dependencies on modperl modules
> here. Hopefully one day the adjusted C::Scan will be
> released and this file will be removed.

OK, thanks - I'll just create a WIN32 constant and use that.

> > Index: lib/Apache/ParseSource.pm
> > ===================================================================
> > RCS file: /home/cvs/modperl-2.0/lib/Apache/ParseSource.pm,v
[ .. ]
> > -                               follow => 1,
> > +                               (Apache::Build::WIN32 ? '' : follow => 1),
[ ... ]

> but isn't that a bug in File::Find? It's supposed to be a
> cross-platform module.

I'll also change that not to use Apache::Build::WIN32. As
for it being a bug in File::Find, it seems reasonable that
File::Find should handle this, but perlport reports that
the -l test for being a symlink is unreliable on Win32
(as opposed to not being implemented).

> > However, with these changes, build/source_scan dies at a
> > certain point:
[ .. ]
> I can't quite help you here Randy, the C::Scan code is
> huge spaghetti of continuous regular expressions. When I
> fixed it to work on linux, I had to inserts some 10,000
> debug statements and try to fix it, which I eventually
> succeeded for the Apache 2.0 source code. I have left some
> of these debug statements commented out. I used Carp to
> get the trace.
>
> > So the problem seem to be the presence of __cdecl in the
> > declaration. I'm afraid I couldn't follow the code around
> > line 809 of lib/ModPerl/CScan.pm enough to know if __cdecl
> > could just be tested for, and thrown away if found, or
> > if it would form part of the declaration needed later?
>
> Try to throw it away and see if it causes any harm, i.e.
> if you get the same xs/tables, then probably no harm was
> done. I could be wrong.

Thanks - I'll try that ...

-- 
best regards,
randy


---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@perl.apache.org
For additional commands, e-mail: dev-help@perl.apache.org


Mime
View raw message