It is in the part of the scripts that were testing what flags to use
for each compiler to use for shared libraries, and assumed that if it
called the compiler with a simple program with -KPIC and got any error
or warning messages, then -KPIC was not correct. That assumption seems
reasonable on the surface, but obviously fails if any other flags are
being passed at the same time that generate warnings as well.
It appears we're not the only ones to see the fragility of this, hence
the improvements in later versions not assuming any warning is a failure.
It does seem strange that Studio 12 is suddenly deprecating flags that
were recommended/required in Studio 11, without any transition period,
but since we have a way to work around it, I'm not going to file any bugs.
(Just don't be surprised if many of your users download existing open
source packages and find they fail to compile 64-bit until they switch
to -m64.)
-Alan Coopersmith- [EMAIL PROTECTED]
Sun Microsystems, Inc. - X Window System Engineering
Chris Quenelle wrote:
Can you say whether such config scripts are broken because they
choose to fail if there are any warnings? Or is because they
are implemented in such a way that they are incapable of ignoring
warnings?
--chris
Alan Coopersmith wrote:
Brian Gupta wrote:
Being that it was released June 4th I'd imagine it's not ready. How
soon do you think that we will move to v12?
Each consolidation is evaluating and testing their own code.
For X, we know Studio 12 doesn't work with our current code base, because
the new "-xarch=generic64 is deprecated" warning breaks configure scripts
built with the old version of the GNU autotools, and switching to -m64
would break builds with Studio 10 & 11. I'm working to find all those
and get their configure scripts replaced with newer ones, but it will
take a little while to do that and then test the results.
_______________________________________________
tools-discuss mailing list
tools-discuss@opensolaris.org
