[PATCH][libobjc,rs6000,testsuite] Fix DDmode ICE on testcase.

Andrew Pinski pinskia@gmail.com
Thu Sep 27 09:17:00 GMT 2007


On 9/26/07, Peter Bergner <bergner@vnet.ibm.com> wrote:
> I wasn't sure what values I could be comparing against, so I went
> with the "safe" option of defining DDmode the same as DFmode so
> we just end up re-calculating the same condition.  If you know
> that changing DDmode to -1 is always safe (ie, never matches),
> I can change that.

-1 will never match as what is being compared against is a character
for the encoding of the type.

> Well, AFAIK, DFP is only defined for C at the moment.
> Do we really want to add DFP support to Objective-C before
> there's a TR or standard document specifying what that
> support will entail?

Objective-C is not really standardized :).  Anyways it is a strict
superset of C (unlike C++) so dfp should just work. The encoding issue
is the only issue with dfp support for Objective-C really, every thing
else should just work, in fact what should happen for the C testsuite
is also run it in Objective-C mode to make sure everything just works.
In fact the parser for Objective-C is just the C parser with some extensions.

If you don't want to file a bug, I will go and file one for dfp and
also the new fixed point type encoding (there is already one for
incorrectly encoding enums and also for vectors).

Thanks,
Andrew Pinski



More information about the Gcc-patches mailing list