PATCH Fix for issue 40411

Joseph S. Myers joseph@codesourcery.com
Mon Jun 7 20:25:00 GMT 2010


On Mon, 7 Jun 2010, Rainer Orth wrote:

> I've now come to the conclusion that this approach isn't appropriate
> here: the conditions which select those files are more complicated than
> what can easily be expressed with pattern matching in specs, and feels
> quite fragile as well.  I'll therefore will introduce two new
> Solaris-specific specs functions %:values-X() and %:values-xpg() to
> return the correct file for the compilation at hand.  At first I thought
> I'd try and introduce some more generic function like %:isoc99() or
> something, but the exact semantics of those files is highly
> Solaris-specific, so that solutions seems best.

Specs functions do seem reasonable here.  FWIW I expect my current 
options/multilibs work will introduce a more structured notion of option 
aliases in both the driver and the compilers proper so that it's no longer 
necessary to check for multiple spellings of an option in cases like this 
(before introducing a proper notion of features to the driver that specs 
can check rather than just option text).

-- 
Joseph S. Myers
joseph@codesourcery.com



More information about the Gcc-patches mailing list