[MacRuby-devel] HAVE_RUBY_VM_H and nkf.bundle (was: installing tmail)

[Laurent Sansonetti]

On May 12, 2010, at 12:42 PM, Laurent Sansonetti wrote:
> Hi Brian,
> 
> On May 12, 2010, at 10:17 AM, Brian Marick wrote:
> 
>> The problem with installing TMail are lines like this in tmailscanner.c
>> 
>> #ifdef HAVE_RUBY_VM_H
>> #include "ruby/re.h"
>> #include "ruby/encoding.h"
>> #else
>> #include "re.h"
>> #endif
>> 
>> I suspect "HAVE_RUBY_VM_H" is being used as a synonym for "on 1.9". It's defined once in the 1.9 source I have, never used. It's not defined in either 1.8.7 or MacRuby.
>> 
>> Is there a more correct way to say "compiling against 1.9"? I don't see anything obvious in the include files, other than parsing it out of paths.
> 
> (I should have read all the messages first :-).)
> 
> I agree, it sounds like they use HAVE_RUBY_VM_H to determine if they build against 1.9. We need to define that variable in MacRuby too.
> 
> I don't think there is a better way yet. I know that Nokogiri does something even worse (it assumes it's 1.9 if the encoding.h header exists, or something similar).
> 
>> Hacking around that gets tmailscanner.c to compile, but tmail also requires 'nkf.bundle', which exists in 1.8 but not in MacRuby. I might try to compile the version out of 1.9, but this pile of yak hair is getting pretty deep, considering I just want to parse RFC822 headers.
> 
> We don't compile the nkf extension yet because we thought it wouldn't been necessary with the new 1.9 encodings API. But it looks like we might have to ship it nevertheless. I suspect we should be able to make a pure Ruby version for it.
> 
> In the meantime, I'm afraid you will have to hack the tmail sources to avoid using nkf (or create a fake nkf.rb file and put it in the path).

I added HAVE_RUBY_VM_H in r4086 and filed https://www.macruby.org/trac/ticket/708 to track the nkf support.

Laurent