r198885 - Revert "Disable LeakSanitizer in TableGen binaries, see PR18325"
chandlerc at google.com
Thu Jan 9 14:48:48 PST 2014
On Thu, Jan 9, 2014 at 2:41 PM, Richard Smith <richard at metafoo.co.uk> wrote:
> On Thu, Jan 9, 2014 at 2:29 PM, Aaron Ballman <aaron at aaronballman.com>wrote:
>> If we're going to back out the revert, can we put the code into an
>> #ifdef so that the reserved namespace identifier is protected when
>> compiling with something that doesn't understand lsan? Then we can
>> argue over the "right" way with some protection.
> I'm not opposed to that, if we have a suitable predefine. But we already
> have *loads* of code in Clang that defines identifiers in the reserved
> namespace (try grepping for '[A-Za-z]__[A-Za-z]' in include/ to find a
> bunch of them), and none of our supported C++ implementations (for clang
> 3.5) have a problem with this, so I don't see that there's a lot of value
> in doing so.
Further, I don't think we should slow down the efforts to get LSan
bootstrapping effectively while we figure out the correct predefine -- we
can add one later as the discussion converges. That had been my plan from
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the cfe-commits