Re: -fpic/-fPIC testresults under sol2.9 & ppclinux trunk

["Kaveh R. Ghazi" <ghazi at caip dot rutgers dot edu>]

 > From: Loren James Rittle <rittle@latour.rsch.comm.mot.com>
 > 
 > In article <200305142257.SAA04943@caip.rutgers.edu> you write:
 > 
 > > Ben - given these facts, would you please take a look?
 > 
 > Hi Kaveh,
 > 
 > Yes, I'm not *Benjamin* but...  I posted the definitive short test
 > case on Monday that displays it is indeed PCH not anything in the C++
 > library itself (sorry, if I failed to CC you).  As Benjamin and I are
 > mainly C++ library guys not PCH guys, I think the ball is in PCH/C++
 > compiler/Geoff's court now...  I have an open PR, other/10757, based
 > on your reports to the libstdc++ mailing list.  Sure, we could paper
 > over it by disabling PCH or perhaps generating more PCH files (in my
 > environment, the bug disappeared when I added extra PCH files); but I
 > think they might want to fix this ICE...
 > 
 > BTW, this is a bug I can find and fix myself, but I know that an
 > expert in that area of the compiler could nail it much faster than I
 > thus I'd prefer to let them do it.  Is this OK?
 > 
 > Regards,
 > Loren

Sorry, I missed your excellent reduction.

It's not my intention to paper over any bug by disabling PCH.  But
running with every test failing because of this problem will mask real
libstdc++ regressions should an unrelated regression occur.  Testing
the library isn't very useful in my current situation...

I'm going to run my personal tests with PCH disabled because of this
until the PR is fixed.  I would suggest we do that globally in the
repo for the same reasons but I won't insist since not everyone runs
multi-passes.

I've added myself to the CC list of the PR.  BTW I didn't see Geoff
CC'ed in the PR, does he know about it?

		Thanks,
		--Kaveh
--
Kaveh R. Ghazi			ghazi@caip.rutgers.edu