Any chance of putting [Inline::SWIG] on CPAN? If there is interest. I was planning to wait.
Thinking of this -0.00 version as an alpha pre-release, to catch newborn infant issues, and determine popular interest. More fundamentally, its parasitation of SWIG (and Inline::C) is rather fragile, so the version's life expectancy is not large. I would be surprised (unhappiliy so) if it were still working a year from now. So CPAN seemed a later step. I'm wondering, is it abstract enough so that we could just make it part of Inline? We could let people pick what type of glue code they want. [Architecture discussion taken offline...] Cheers, Mitchell Date: Wed, 2 Oct 2002 00:03:17 -0700 From: Brian Ingerson <[EMAIL PROTECTED]> To: Mitchell N Charity <[EMAIL PROTECTED]> Cc: [EMAIL PROTECTED], [EMAIL PROTECTED] Subject: Re: Inline::SWIG Message-ID: <[EMAIL PROTECTED]> References: <[EMAIL PROTECTED]> In-Reply-To: <[EMAIL PROTECTED]>; from [EMAIL PROTECTED] on Tue, Oct 01, 2002 at 12:16:26AM -0400 On 01/10/02 00:16 -0400, Mitchell N Charity wrote: > SWIG (http://www.swig.org/) does some nice parsing and glue generation. > In particular, it has rather broader coverage of C++ than Inline::CPP > (templates, operators, etc). > > Perl's Inline (http://www.perldoc.com/cpan/Inline.html) is sometimes > more convenient. Particularly for small bits of code. Glue generation, > compilation, caching, and linking, all happen automagically. And it > facilitates run-time code generation. > > So I wrote an Inline::SWIG > http://www.vendian.org/mncharity/dir3/inline/swig/ Mitchell, This sounds awesome. You're a genius. Any chance of putting it on CPAN? I could help you there. I'm wondering, is it abstract enough so that we could just make it part of Inline? We could let people pick what type of glue code they want. I was thinking of writing my own glue code, sans XS, so that we could get rid of XS altogether. We could do it without Makefile.PL too which I think would make code generation/compiling very fast. Just: - Parse using your regex parser (instead of RecDescent) - Generate pure C glue - Feed it right to gcc So now we'd have three different gluing options - SWIG - XS - IngyLine Don't worry folks, I won't throw out the XS for good. It's way too stable. But a faster option might be good for some applications. Cheers, Brian
