On Mon, Feb 20, 2012 at 11:13:33AM -0800, Chandler Carruth wrote:
> On Mon, Feb 20, 2012 at 8:39 AM, Joerg Sonnenberger <[email protected]
> > wrote:
> 
> > Hi all,
> > this is the minimally intrusive fix for the target-prefixed toolchain
> > support. This is a partial revert of r148839, fixed in r149083, reverted
> > and re-broken in r149337. Given that's been 3 weeks now with a proper
> > solution, I think it is time to restore the status quo. See also PR
> > 12026.
> >
> 
> This is a hack though, and it's important to acknowledge that. We're now
> handling -target in two separate places. It would be better to sink this
> down to be grouped with the 'getToolChain()' call that is relevant, and add
> comments at the very least.

Like I said, this is just the minimal version to get it back working.

> Also, to introduce a hack like this, we definitely need good test cases so
> that we understand what is and isn't supposed to work. Relying on the
> happenstance of DefaultTargetTriple is what got us into the mess in the
> first place.

The test cases can be revived from r149083.

Joerg
_______________________________________________
cfe-commits mailing list
[email protected]
http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits

Reply via email to