Re: [PATCH GOLD] [6/N mingw host] Add support for not keeping the files open on execute

[Viktor Kutuzov <vkutuzov at accesssoftek dot com>]

> > #if defined(_WIN32) && !defined(__CYGWIN__)
>
>  Any reason why not just "#ifdef __MINGW32__" for that case?

This is some kind of a pattern which is used all over the places, so, people would expect to see something like this.
However, I'm not sure about _WIN32. Do we target 64-bit Windows at all?

Cheers,
Viktor

----- Original Message ----- 
From: "Dave Korn" <dave.korn.cygwin@googlemail.com>
To: <gcgb-binutils@lo.gmane.org>
Cc: "Andrew Pinski" <Andrew_Pinski@playstation.sony.com>; "binutils" <binutils@sourceware.org>
Sent: Wednesday, October 07, 2009 5:35 PM
Subject: Re: [PATCH GOLD] [6/N mingw host] Add support for not keeping the files open on execute


> gcgb-binutils  wrote:
> > Andrew Pinski wrote:
> > > This is the first patch which I could not figure out how to do without
> > > a #ifdef _WIN32.
> >
> > This may break cygwin (that is, it may unintentionally cause a gold
> > built on a cygwin $host to use win32 constructs directly, instead of
> > going thru the cygwin posix emulation).
>
>  Oh, good catch!
>
> >    If you really truly want a block of code
> > to be active ONLY if mingw, then it's best to use:
> >
> > #if defined(_WIN32) && !defined(__CYGWIN__)
>
>  Any reason why not just "#ifdef __MINGW32__" for that case?
>
> > And, if you want a block of code to be active on the windows platform,
> > regardless of whether its mingw or cygwin, then
> >
> > #if defined(_WIN32) || defined(__CYGWIN__)
>
>  Likewise.  After all, that one will theoretically also catch MSVC, CE and
> Interix, won't it?  :-P
>
>
>    cheers,
>      DaveK