This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: Death to configure.frag


On Fri, Aug 29, 2003 at 04:25:41PM +0530, Ranjit Mathew wrote:
>Zack Weinberg wrote:
>>>>The odd-looking idiom 'include /dev/null $(variable)' prevents make
>>>>from grousing if $(variable) expands to nothing.  No, -include does
>>>>not prevent that error, I tried it.
>>>
>>>What I use is this:
>>>
>>>ifneq ($(variable),)
>>>include $(variable)
>>>endif
>>>
>>>On some OSs, opening files is expensive.  On DJGPP, for example, the
>>>entire /dev directory is simulated.
>>>
>>>In this case, however, it's easier to have those variables default to
>>>/dev/null in configure.
>> 
>> 
>>Actually, no, it isn't, and I'm a little worried about the existence of
>>/dev/null on (e.g.) Mingwin anyway.  I will revise the patch to use the
>>ifneq idiom you suggest above.
>
>I think you mean MinGW, in which case, GCC builds are normally done on
>MSYS (http://www.mingw.org/msys.shtml), a UN*X environment over Windows
>which has no problems with "/dev/null".
>
>If one is using CygWin on Windows, this is even less of a problem, if
>at all.

Just to set the record straight, MSYS is essentially cygwin with some
hacks, despite the fact that cygwin isn't mentioned on the above page.

cgf


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]