[ELDK] Rebuilding python ELDT package

[Vitaly Bordug]

В Sat, 4 Jul 2009 11:56:03 +0200
Matthias Fuchs <mfuchs at ma-fu.de> пишет:

> Hi,
> 
> I am trying to build - modify - build again the ELDT python RPM
> package.
> 
> I am using the latest tarball and spec file from the Denx git
> repositories.
> 
> The initial build as described on
> http://www.denx.de/wiki/view/DULG/ELDKRebuildingComponents works
> fine. Good work.
> 
> But now I want to modify the python source and rebuild - but without
> doing the complete rpmbuild. Because I am not an RPM expert I
> google'd and found this recipe - still thinking this should be the
> same as "rpmbuild -ba"
> 

no need to be RPM expert. Just the other ways are not well tested, so
you may approach "easy" way with that.

> 1) Prepare the sources (unpack tarball etc.):
> ppc_4xx-rpmbuild -bp python.spec
> 
> 2) Modify sources (or not)
> 
> 3) start at build stage
> ppc_4xx-rpmbuild -bc --short-circuit python.spec
> 
> But now I get:
> 
> b.h: found (4, 5) in /opt/eldk_420/ppc_4xx/usr/include
> db lib: using (4, 5) db-4.5
> checking build system type... i686-pc-linux-gnu
> checking host system type... i686-pc-linux-gnu
> checking target system type... i686-pc-linux-gnu
> checking for gcc... ppc-linux-gcc
> checking for C compiler default output file name... a.out
> checking whether the C compiler works... configure: error: cannot run
> C compiled programs. If you meant to cross compile, use `--host'.
> See `config.log' for more details.
> Failed to configure _ctypes module
> /opt/eldk_420/usr/src/denx/BUILD/Python-2.5.1/Modules/readline.c: In
> function
> 'flex_complete': /opt/eldk_420/usr/src/denx/BUILD/Python-2.5.1/Modules/readline.c:681:
> warning: return makes pointer from integer without a cast
> + exit 0
> 
> Q1: Is there any difference between 
> 	ppc_4xx-rpmbuild -ba python.spec
> 
> and 
> 
> 	ppc_4xx-rpmbuild -bp python.spec; ppc_4xx-rpmbuild -bc
> --short-circuit python.spec ?
> 

Rhetorical question -you just showed they are different. IMHO, cross
rpmbuild is tricky enough, and having many params that involve macros
etc. may lead to confusion.

> 
> Q2: I noticed that during the complete build (-ba) I always get two
> configure runs (hostpython + target's python). This is correct. But
> the spec file seems to take take of the case when hostpython and
> hostpgen already exist from a previous build. This check does not
> work. I had to modify the spec file to use $topdir instead of
> $RPM_BUILD_DIR:
> 
> ...
> if [ ! -e $topdir/hostpython -a ! -e $topdir/Parser/hostpgen ]; then \
>     if [ -e $topdir/Makefile ]; then $(MAKE) -C $topdir distclean;
> fi; \ ...
> 
> Am I missing something or is this a bug in the spec file?
> 
It seems to me you have rpm macros confusing itself. python build worked
the way it is to me, last time I had a look. Nevertheless I'll try to
reproduce your problem in a little while.

Sincerely,
	Vitaly
> 
> Matthias
> 
> _______________________________________________
> eldk mailing list
> eldk at lists.denx.de
> http://lists.denx.de/mailman/listinfo/eldk