Re: egcs-1.1 status (PPC)

[Franz Sirl <Franz dot Sirl-kernel at lauterbach dot com>]

Am Wed, 19 Aug 1998 schrieb Jeffrey A Law:
>In message <199808181147.EAA14980@cygnus.com>you write:

>  > I haven't seen this bug in the kernel yet, most "bugs" I see now reported
>  > are in complex user programs (2 so far, one of them you just fixed,
>  > thanks!) or due to the stricter C++ handling. This is really a big step
>  > forward for egcs on PPC, I even saw the first reports complaining about
>  > compiler performance :-) and no longer about compiler bugs.
>Cool.  We've still got the PIC register problem (980526-1), which
>Meissner will have to help with (fixing it looks to be rather complex),
>varargs problems, and a bug or two that you've recently reported.

And the bug reported by Gary seems to be fixed already (probably the change to
haifa-sched.c). But is the PIC problem really that difficult? I thought the
difference between -fpic and -fPIC wasn't that big, and -fPIC works with the
testcase.

>  > That would be fine. But I can tell you that the current cvs-glibc compiled
>  > fine yesterday and works without problems on my machine so far. I haven't
>  > run "make check" though ;-).
>The magic tag is "glibc-4-egcs-1_1".  That's the version that needs
>to successfully build and check.  If you could verify that for the
>ppc, it would be greatly appreciated.. 

Sure, I will test it. I saw Geoff already reported a problem with glibc and
included a fix.

>  > Other progs: 
>  > - KDE-1.0 compiles and works (kdegames/ksnake fails because of bad c++ sour
>  > ce)
>  > - qt-1.33 compiles and works
>  > - qt-1.40 compiles, but has problems with drag-select, not investigated yet
>  > - mozilla: no status yet
>  > - jdk: no status yet
>  > - ssl-0.9.0b: one bug report I have to investigate yet
>This is excellent information to get.  While I won't hold up the release
>for any of these things, we do benefit by trying to build them and
>fixing what bugs we can (or noting that it did successfully build).

These applications are likely to be compiled by a lot of unexperienced users
and in order to avoid repeated bug reports I try to make sure that they at
least compile.

I just tried ssl myself and it passes all tests. I will try to
contact the reporter for his setup.

Franz.