0) Problem: Some people have an additional link to the python man-page:
is the python footprint mismatch something I can fix or am I missing anything important?
NEW lrwxrwxrwx root/root usr/man/man1/python.1.gz -> python2.1.gz
1) the relevant part of Makefile.pre.in:
maninstall: altmaninstall
-rm -f $(DESTDIR)$(MANDIR)/man1/python2.1
(cd $(DESTDIR)$(MANDIR)/man1; $(LN) -s python$(VERSION).1 python2.1)
-rm -f $(DESTDIR)$(MANDIR)/man1/python.1
(cd $(DESTDIR)$(MANDIR)/man1; $(LN) -s python2.1 python.1)
2) after 'make install' we have the following chain of symlinks:
lrwxrwxrwx 1 juergen users 9 Apr 21 09:13 python.1 -> python2.1
lrwxrwxrwx 1 juergen users 11 Apr 21 09:13 python2.1 -> python2.7.1
-rw-r--r-- 1 juergen users 14582 Apr 21 09:12 python2.7.1
3) the compress_manpages function of our pkgmk:
compress_manpages() {
local FILE DIR TARGET
cd $PKG
find . -type f -path "*/man/man*/*" | while read FILE; do
if [ "$FILE" = "${FILE%%.gz}" ]; then
gzip -9 "$FILE"
fi
done
find . -type l -path "*/man/man*/*" | while read FILE; do
TARGET=`readlink -n "$FILE"`
TARGET="${TARGET##*/}"
TARGET="${TARGET%%.gz}.gz"
rm -f "$FILE"
FILE="${FILE%%.gz}.gz"
DIR=`dirname "$FILE"`
if [ -e "$DIR/$TARGET" ]; then
ln -sf "$TARGET" "$FILE"
fi
done
}
4) How it works:
We search for real man-pages in $PKG and compress them, breaking all
symlinks to these man-pages, which we repair afterwards. Fixing is done by
adding a .gz extension to all targets of the symlinks.
However, we do a test if $TARGET exists before we do the actual link,
which is not the case if we have chained symlink and we do not process
in the right order.
For a "normal" collating we have the order python.1 before python2.1, so
python2.1.gz does not exits at the time we process python.1
