From: Luis Lavena <luislavena@...>
Date: 2010-11-09T03:06:33+09:00
Subject: Re: Using Ruby's C API safely and sanely

On Nov 8, 1:45 pm, Eric MSP Veith <eve...@wwweb-library.net> wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> Hello Rubyists!
>
> I'm trying to integrate Ruby (both for bindings and embedded interpreter)
> into a larger project, namely RPM 5. I'm facing several diffuclties there
> and hope that somebody might point out a more elegant way of solving things.
>
> First, querying Ruby's compile-time configuration is rather painful. RPM 5
> uses Autoconf, and using extconf.rb as yet another layer of configuration
> seems not the best way. Currently, we're using calls to "ruby -rrbconfig -e
> puts CONFIG::expand ..." to get information like CFLAGS/include directories,
> location of Ruby's library directories, and so on. Is there a better way to
> do this? Sadly, Ruby doesn't come with a .pc file or something similar, and
> a call like Perl's "perl -V:..." doesn't seem to be implemented, either?
>
> Second, including the ruby.h file leads to the inclusion of Ruby's config.h
> file. This is clearly not appreciated, as we need to cope with a lot of
> HAVE_FOO_H defines, or even Ruby's PACKAGE_NAME, etc. definitions which of
> course override RPM's. Is there a sane way of using Ruby's C API, i.e.
> including the Ruby header file, without having Ruby wreak havok on our own
> defines?
>
> Thanks alot in advance. Every hint is greatly appreciated.
>

Ruby trunk includes a package configuration file.

You should checkout trunk and use it as base of your project.

--
Luis Lavena