[llvm-commits] [PATCH] Re-introduce llvm::sys::getHostTriple() function.

Kaylor, Andrew andrew.kaylor at intel.com
Tue Jan 15 11:40:49 PST 2013


I don't think it's good to have the 32-on-64 test cases broken on Windows.  I believe at least one buildbot uses that combination.

Is there a reason that the tests weren't acceptable the way they were (i.e. using the target triple everywhere)?

-Andy

-----Original Message-----
From: Peter Collingbourne [mailto:peter at pcc.me.uk] 
Sent: Monday, January 14, 2013 6:09 PM
To: Kaylor, Andrew
Cc: llvm-commits at cs.uiuc.edu; spop at codeaurora.org; joerg at NetBSD.org
Subject: Re: [llvm-commits] [PATCH] Re-introduce llvm::sys::getHostTriple() function.

On Fri, Jan 11, 2013 at 10:23:29PM +0000, Kaylor, Andrew wrote:
> Hi Peter,
> 
> Sorry to take so long to get back to you about this.
> 
> I just tried out your patch and it does not work on Windows.  It seems that 'sed' is required to build 'llvm-config' so 'llvm-config' is not built on Windows.

I switched to using the host triple for the Windows checks in lit.cfg.
This will mean that 32-on-64 won't work properly on Windows, but I can't afford to invest any more time fixing this on Windows.
I managed to get access to a Windows machine; this works fine on MSYS, haven't tried MSVC.

Thanks,
--
Peter




More information about the llvm-commits mailing list