Re: brute force builder state
On Wed, Mar 22, 2000 at 10:50:53AM +0100, Guido Guenther wrote:
> On Tue, Mar 21, 2000 at 07:51:22PM +0100, Florian Lohoff wrote:
> > Hi,
> > Most of them failed due to kernel header problems - Probably i should compile
> > against the 2.2.1 kernel sources - I have now CVS Snapshot 2.3.51 installed.
> The 2.2 kernel headers give you the same SOCK_* undefined messages. I just commented
> out the "#ifdef __KERNEL__" around these definitions at the end of /u/i/asm/socket.h.
> This works well for me(with a 2.2 Kernel but 2.3 should be as good as
> that).
This was the pointer i needed - Ill try this - There was a big discussion
on the SGI Linux Mips list which was not THAT interesting and they did not
come up with a correct solution.
> > The second big failure reason is direkt or indirekt missing xfree like
> > libgtk, groff, gnome, xmkmf.
> Did you try to build a xlib6? I had a short look at it but the package
> is rather big and confusing ;).
Nope - This will be one of the things ill take care of when i am finished
with a complete build run i think - It takes some Aehm - hours to compile
on my machine :)
> > clisp none linker error
> These linker errors worry me a lot. Any idea what causes them?
Linker problems :) - Honestly - The binutils for mips(el) are the most
ugliest things i have been told - There are a couple of people working
on it to fix these linker errors and add symbol versioning (without this
the glibc 2.2 wont be possible)
Most of the linker errors can be worked around by trying different
revisions of binutils - I had much luck with 991114 or something which
did compile most of the things 2.8.1 + mips patches didnt survive.
Just try to use current CVS binutils - Sometimes it works :)
Flo
--
Florian Lohoff flo@rfc822.org +49-5241-470566
"Technology is a constant battle between manufacturers producing bigger and
more idiot-proof systems and nature producing bigger and better idiots."
Reply to: