[ELDK] Problems rebuilding ELDK-4.1 from scratch (RCS keyword expansion)

[J.T. Conklin]

Hi,

On Wed, Mar 18, 2009 at 4:43 PM, Wolfgang Denk <wd at denx.de> wrote:
>> With the workspace I checked out this morning (with git-clone, then
>> git-checkout ELDK_4_1), I get these instances where the RCS Id's are
>> not expanded:
>>
>> $ find . -name '*.patch' | sort | xargs fgrep '$Id$'
>> ./cross_rpms/crosstool/SOURCES/uclibc-0.9.27-update-math.patch:- * $Id$
>> ./cross_rpms/crosstool/SOURCES/uclibc-0.9.27-update-math.patch:+ * $Id$
>> ./cross_rpms/rpm/SOURCES/db-3.1.17-jbj.patch:+ * $Id$
>> ./target_rpms/crosstool/SOURCES/uclibc-0.9.27-update-math.patch:- * $Id$
>> ./target_rpms/crosstool/SOURCES/uclibc-0.9.27-update-math.patch:+ * $Id$
>> ./target_rpms/mtd_utils/SOURCES/mtd_utils-2.6.19.patch: /* $Id$ */
>> ./target_rpms/rsh/SOURCES/netkit-rsh-0.16-jbj2.patch:-  "$Id$";
>> ./target_rpms/rsh/SOURCES/netkit-rsh-0.16-jbj2.patch:+  "$Id$";
>
> I still don't understand why you think these strings should be any
> kind of problem?

These patches are applied to SRPMs.  But when the SRPMs are unpacked,
those source files have RCS Id's with contents (like $Id: ... $)
instead of $Id$,  and those won't match what's expected by the patch.
So the patch utility will fail to apply the patch, and then the build
stops with an error.

    --jtc