[cfe-dev] PATCH - Support for c++ includes directories mingw-w64

[Gordon Keiser]

>> From: cfe-dev-bounces at cs.uiuc.edu [mailto:cfe-dev-bounces at cs.uiuc.edu] On Behalf Of Chandler Carruth
>> Sent: Saturday, October 06, 2012 11:00 PM
>> To: Rafael Espíndola
>> Cc: Christer Solskogen; clang-dev Developers
>> Subject: Re: [cfe-dev] PATCH - Support for c++ includes directories mingw-w64
>> 
>> On Sat, Oct 6, 2012 at 9:39 AM, Rafael Espíndola <rafael.espindola at gmail.com> wrote:
>> > I have no idea what you are talking about. This is my very first patch
>> > to clang, and it just solved my problem. If there are better places to
>> > patch to achieve the same goal, I really don't care :-)
>> There is a new way that is used by linux, but I don't think anyone has
>> signed up for moving mingw to it.
>> 
>> Chandler, do you think Christer's patch is OK?
> 
> Not that there is anything wrong with the patch, but we really shouldn't be adding hacks here. If someone cares about mingw header search, the patch sholud be to move that logic to the driver and fix it there.
> 

I suspect that adding any more hacks to the header / lib searching parts of the driver may cause it to collapse into a singularity. 

Seriously though, I've been wanting to refactor this part of Clang for a while.  At the very least getting the various platforms moved into their own files would probably make a big difference in readability.  I just have to find a slow enough period at work, or get the time allocated for it specifically.

> 
> Notably, doing this will remove the need for what Christer's patch is doing -- the driver has logic to search for any viable GCC version in the header search path rather than hard-coding a list. I think that's the proper way to fix this.
>  
> 
> > --
> > chs,
> 

Cheers,
-Gordon