Charles Wilson wrote:
If THAT doesn't exercise the mixed-mode patch, nothin' will.
Except that it also uses the src_prep_init_hook changes. Which only
makes sense: if you download third-party or upstream patches, somehow
you've got to "do" something with them. PATCH_URI helps, but in this
case the lossless-jpeg "patch" contains binary images that must be
copied into src/ somehow. And that's what the jpeg.cygport does in
src_prep_init_hook.
I could probably simplify jpeg.cygport's src_prep_init_hook() a little
bit by putting
${origsrcdir}/ljpeg-6b.patch
${origsrcdir}/100_crop.patch
${origsrcdir}/200_crop_man.patch
${origsrcdir}/201_rdjpgcom_locale.patch
${origsrcdir}/202_jpeglib.h_c++.patch
${origsrcdir}/203_rdppm.patch
${origsrcdir}/204_exif_extra.patch
into PATCH_URI, but that still leaves the following in src_prep_init_hook():
cd ${origsrcdir}
mv testimgl* ${SRC_DIR}/
cd ${SRC_DIR}
But those images are only needed by src_test().
However, the above procedure doesn't work, because:
(1) the two "extra" tarballs are unpacked into ${origsrcdir}
(2) cygport strips directory specifiers from PATCH_URIs:
for _patch_uri in ${PATCH_URI}
do
_src_orig_patches="${_src_orig_patches} ${_patch_uri##*/}";
done
and for good reason, too! Those PATCH_URIs might have http://
or worse in them.
(3) _src_orig_patches therefore contains "bare" filenames with no
directory, and apply_patch is called as:
for src_patch in ${_src_orig_patches}
do
apply_patch ${top}/${src_patch};
done
I don't see either of those for-loops being changed, because they do
what they do for very good reasons; changing their behavior would break
lots of stuff. Maybe:
for src_patch in ${_src_orig_patches}
do
if [ -e ${origsrcdir}/${src_patch} ]
then
apply_patch ${origsrcdir}/${src_patch}
else
apply_patch ${top}/${src_patch};
fi
done
But what if your third-party tarball contained:
one-patch
two/another-patch
three/yet-one-more-patch
So, even if jpeg.cygport uses PATCH_URI as above, it'd still need a
src_prep_init_hook() function to move the patches from ${origsrcdir} to
${T}. Frankly, I'd prefer to keep them in ${origsrcdir} and call
apply_patch -- for THOSE patches -- explicitly, if I'm gonna have a
src_prep_init_hook() anyway.
When you're ready for another test case for one of the other patches,
let me know.
Sorry, didn't see the next message you sent asking for an demo of the
"hooks" stuff. Obviously the mixed-mode-demo patch will do so for the
src_prep_*_hooks, but I'll send a demo that needs the
src_install_*_hooks in a separate message...
--
Chuck
--
Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple
Problem reports: http://cygwin.com/problems.html
Documentation: http://cygwin.com/docs.html
FAQ: http://cygwin.com/faq/
