Hi,
I was trying to "patch a patch" in a u-boot recipe using standard
bbappends technique. I added a task before do_patch after do_unpack
to munge the patch that was broken. Here is what seems to happen:
do_unpack:
copies all the patch files from the metadata layer into ${WORKDIR}
do_fix_patch (my additional task)
uses sed or patch to apply a local fix to a broken patch in ${WORKDIR}
do_patch
creates a subdirectory in ${S} called "patches", and puts links to
the actual files in the layer and uses those links back to the layer
files to apply the patches.
It seems quilt is using the actual layer files instead of the ones in
${WORKDIR}??
This seems odd and somehow wrong. We copy the layer files into
WORKDIR, and the unsuspecting user (me) thinks those are the files
that do_patch will use.
This one took me some time (and help) to figure out!
Comments? Is this a bug? halfway in the middle of some architectural
changes? Expected behavior?
My workaround was to simply replace the entire patch.
-Chris
--
Life is like Linux - it never stands still.
--
_______________________________________________
yocto mailing list
yocto@yoctoproject.org
https://lists.yoctoproject.org/listinfo/yocto
