On Sun, Mar 17, 2024 at 04:18:35PM +0200, Eli Zaretskii wrote:
> > Date: Sun, 17 Mar 2024 13:54:21 +0100
> > From: <tomas@tuxteam.de>
> > 
> > after a while, I'm trying again to pick up the custom of
> > compiling Emacs.
> > 
> > Usually, I compiled in an empty build directory, invoking
> > ../emacs/configure.
> 
> That is the correct way.

Thanks for confirming.

> > This used to work pretty well but doesn't seem to these
> > days. I hit two-and-a-half snags:
> > 
> > - one missing chartab.o which happened to lie around 
> >   in the source dir under src; copying it over to the
> >   corresponding spot in the build dir let the build
> >   continue
> 
> How come?  chartab.c is just one of the source files, and is not
> mentioned specially in any Makefile, AFAICT.  Why is that file somehow
> missing, and others aren't?  Please investigate why.  (Could it be
> your src tree was not clean of all the *.o files?)

Good question. I was surprised myself -- anyway, copying chartab.o
from <source>/src/ (why was it there in the first place?) to
<build>src/ let the build continue, until...

> > - one missing symbol rpl_chmodat; this gets defined in
> >   some header -- sneaking a strategically placed
> > 
> >     #undef rpl_chmodat
> 
> If this is on GNU/Linux, then you shouldn't need to have any
> rpl_chmodat anywhere.  I just searched my source/build tree on
> GNU/Linux, and didn't find rpl_chmodat anywhere.  So once again,
> please investigate.

It is GNU/Linux, yes, and this was my take too, since "my" libc
does carry a chmodat.

> And in any case, this doesn't seem to have anything to do with
> out-of-tree builds, does it?

Thing is, in-tree build (after a make bootstrap) succeeds right
away. Out of tree builds (also after make bootstrap) failed for
me as described above.

So I'll definitely have a look. Next week is a bit busy, so it
might take me until weekend.

Thanks for confirming that (a) out-of-tree build is supposed to
work and (b) there is interest in knowing when it doesn't.

Cheers
-- 
t