Axalon Bloodstone wrote:
>
> On Sun, 23 Jan 2000, John Cavan wrote:
>
> > Sheesh, people are going to associate me with QA work, and I'm a
> > developer... ;o)
> >
> > Anyways, I found a few more things...
> >
> > 1) ExclusiveArchs for lother-etherconfig and ltrace prevent K6 (and a
> > few other x86 flavours).
>
> You should wait for/bug Dadou, pretty sure K6 is built
Already done myself... in fact, I now have nearly the entire Mandrake
SRPM tree rebuilt for the K6, with a few exceptions (such as egcs).
> > 2) nist-0.6-2 requires nasm to build, but nasm is not a part of the main
> > distro... downloaded the source and built it myself, but maybe it should
> > be included? It's useful, some of the console game emulators use it.
>
> it's in contrib/
I knew that, I was just wondering why it wasn't in the main
distribution. There's another one found in contrib that is required to
build a SRPM (I can't remember which). I was only wondering why it's not
in the man distribution if it is a required piece for building main
distribution software...
> > 3) pmake problem:
>
> this seems intentional :/
Shouldn't be though, the build will fail when pmake uses itself to
configure for building on the second pass. It guesses the architecture
and pulls up i586 (on a K6 as well) and then fails to find the
appropriate file. Anyways, I believe Chmouel is fixing it. :o)
> > cp $RPM_SOURCE_DIR/pmake-sys-${RPM_ARCH}.mk lib/mk/sys-%{_arch}.mk
> >
> > will copy pmake-sys-i386.mk to lib/mk/sys-i386.mk and not to sys-i586.mk
> > which pmake will look for later in the build. I would think that this is
> > a little backwards...
> >
> > cp $RPM_SOURCE_DIR/pmake-sys-%{_arch}.mk lib/mk/sys-%{_target_cpu}.mk
> >
> > Probably require a test for k6 so that the file gets copied to
> > sys-i586.mk since nothing seems to be able to say that the k6 is the
> > current cpu... I've gotta figure out how to change that. :o)
> >
> > 4) util-linux calls:
>
> seems someone forgot to check for changes, make sure you test that before
> you need it after optimizing it's kinda notorious for breaking. :)
So far so good... I'll let you know. The default spec seemed pretty odd
since util-linux definitely knows nothing about RPM! I found the call to
"OPT" by digging through the configuration files and I would guess that
this was the desired behaviour. I chalk those typos up to late nights...
:o)
> > make "RPM_OPT_FLAGS=$RPM_OPT_FLAGS"
> >
> > when it should be calling:
> >
> > make "OPT=$RPM_OPT_FLAGS"
> >
> > That will correctly pass the optimization flags to the compiler,
> > otherwise it's 486 city...
John
