On Sun, Jan 20, 2013 at 07:58:06PM -0800, Paul Rogers wrote:
> I've gone back to follow the BLFS-6.3 book (I know, FBBG) and install
> Xorg-7.2 & Mesa-6.5.2. Got the Mesa tarball from the Mesa site. Oddly,
> the build script copy/pasted from the book falls out after the make, but
> there's no error number set. (???) Looking at the log file my build
> script makes, there are many warnings that some headers can't be found.
> Seems from the names they should be standard, and I have, for example,
> stddef.h, but not where it looks--it's in the c++ includes:
>
> float.h
> limits.h
> stdarg.h
> stddef.h
>
> They're not with the package. I can't check the support archives yet.
> Suggestions, please? TIA
I took a look at my old logs - for LFS-6.6 I was using a much newer
version of Mesa. Then I searched through my notes : I did find a
similar problem, but not a solution.
In 2006 I was trying out modular Xorg, particularly on ppc and
ppc64 in cross-lfs. It looks as if I hit this problem on 64-bit
ppc64 multilib - 32-bit apparently built ok. The video card in that
machine is from nvidia, and I eventually concluded that building
Mesa for the nv driver was not particularly useful.
However, when I looked at my notes from the rest of that year I
remembered how hard it used to be to build Mesa on second-class
architectures, and how there were a large number of different build
targets ('make linux-dri' and variants). I know that you are on
32-bit x86 which was the first-class architecture in those days, but
possibly it is only one of the drivers that fails to build.
ISTR each driver had its own directory, and it looks like sedding
them out of the Makefile was the way to avoid problematic drivers,
but YMMV.
You say you don't have an error number, but does it actually say
'Error:' in the log ? If it doesn't, the missing headers might not
be important. If it does say that, check where it is failing
(which directories and files). Then look at the Makefiles to see if
you think you can usefully (for you) build it without that/those
part(s) of the code.
My view is that you would have done better to accept the results
from your earlier build, because no problem other than an error
message had been identified : maybe in due course it would indeed
lead to a problem, maybe it wouldn't. But it's your system and your
time. The versions in 6.3 are now so old that I wouldn't touch them
with the proverbial barge-pole where newer versions exist. Most
people have not retained records from that long ago, so google is the
only likely source of help. Good luck, whichever approach you take.
ĸen
--
das eine Mal als Tragödie, das andere Mal als Farce
--
http://linuxfromscratch.org/mailman/listinfo/blfs-support
FAQ: http://www.linuxfromscratch.org/blfs/faq.html
Unsubscribe: See the above information page
