Re: doc-base is hugely unloved; bug mass-filing needed?

[Ian Zimmerman <itz@buug.org>]

Stefan Fritsch:

Stefan> This version produces false positives. I suspect it's because
Stefan> you look at all file names and not just those under
Stefan> /usr/share/doc

Stefan> apache2.2-common

Stefan> This does not contain any docs (but it contains html files). The
Stefan> docs are in apache2-doc and are registered with doc-base.

Fixing that is not easy (at least in a shell script), because too many
packages contain symlinks from within /usr/share/doc to somewhere else:

#! /bin/bash

outside_docdir()
{
    local d="$1"
    local reald=`readlink -f "$d"`
    case "$reald" in
        /usr/share/doc/*)
            ;;
        *)
            echo "$reald"
            ;;
    esac
}

find -L /usr/share/doc -name "*.html" -o -name "*.xhtml" -o -name "*.xhtml.gz" \
    -o -name "*.pdf" -o -name "*.pdf.gz" | \
    while read d ; do outside_docdir "$d" ; done | \
    xargs dpkg -S | cut -d: -f1 | sort | uniq

yields:

gap-character-tables
gap-doc
gap-matrix-schreiersims
gimp-help-en
gimp-help-fr
libatk1.0-doc
libgail-doc
libglade2-dev
libvte-doc
pari-doc
python-gtk2-doc
sgt-puzzles
xorg-docs


IIRC I filed a bug about this too, in the case of the monstrous
/usr/share/gtk-doc directory, but nothing ever came out of it.

It is easier to handle this in perl or python, with a
hash/dictionary/set of the symlinks.  But I'm starting to fear I'm
wasting my time and effort here. :-(


-- 
Ian Zimmerman <itz@buug.org>
gpg public key: 1024D/C6FF61AD 
fingerprint: 66DC D68F 5C1B 4D71 2EE5  BD03 8A00 786C C6FF 61AD
Ham is for reading, not for eating.