Re: Systemd Multilib build success

Am Sonntag, dem 22.12.2024 um 13:47 +0800 schrieb Xi Ruoyao:
> On Fri, 2024-12-20 at 16:24 +0100, Thomas Trepl wrote:
> > Am Dienstag, dem 17.12.2024 um 20:52 -0500 schrieb Eric Clothier:
> > > I wanted to report that I was able to successfully build MLFS using
> > > jhalfs with systemd init. I have NOT done any major testing of the
> > > build, but it does successfully build and boot. I did test an ml_32
> > > build only, as my host distro does not include the X32_ABI kernel
> > > flag, and I simply have not taken the time to build a kernel for my
> > > host distro from source.
> 
> But I can see some obviously questionable things in the systemd
> instructions:
> 
> - Why do we need a xsltproc symlink?  The build system should disable
> building man pages when xsltproc is not available.
> - The instruction seems installing "/usr/lib32/systemd/*" which are
> completely meaningless.  We only want libudev.so* and libsystemd.so*.
> - 'CFLAGS+=" -Wno-error=shift-overflow"' seems covering up a real bug,
> instead of covering up it you should really fix the bug and preferably
> send the fix upstream.
> - Why "-march=i686" for lib32?  We don't add it for all other packages.
> 
> And maybe more.  The items above are just found in a quick one-minute
> review.
> 
> Thus IMO it's broken.  If it happens to work for you, it just happens.
Obviously, its not that broken as users do use the instructions to build
their system successfully. Of course, the instructions might be buggy - in
this case, a patch or so is appreciated. As i said, i maintain the systemd
variant on a "best guess" basis. I'll be happy if someone sends me patches
if there is something to fix.
Maybe i should note that on the MLFS page, too.

-- 
Thomas

-- 
http://lists.linuxfromscratch.org/sympa/info/alfs-discuss
Unsubscribe: See the above information page