Re: Repeated files in -doc packages

[Iain Lane]

Hey,

On Sun, Dec 20, 2009 at 10:20:36AM -0200, Marco Túlio Gontijo e Silva wrote:

Hi.

I noticed all -doc packages include some repeated files: plus.gif,
haddock-util.js, haskell-icon.gif, minus.gif and haddock.css.

In ghc6-doc there're a lot of them, as can be seen in
http://packages.debian.org/search?searchon=contents&keywords=haddock-util.js&mode=path&suite=unstable&arch=any
 .

I've made this simple script to check if the files have the same
content, and they have:

#!/bin/sh

set -e

[ "$1" = "" ] && echo USAGE: $0 FILE && false

files=`dpkg -S $1 | grep -e ghc -e haskell | cut -d\  -f2`
nfiles=`echo $files | wc -w`

file()
{
   echo $files | cut -d\  -f$1
}

for i in `seq $nfiles`
do
   for j in `seq $((i + 1)) $nfiles`
   do
       diff -q `file $i` `file $j`
   done
done

Should we do something about it?

Greetings.

Cheers,
Iain

    [ -n "$$CDBS_NO_DOC_SYMLINKING" ] || \
    [ -h debian/$(cdbs_curpkg)/usr/share/doc ] || \
    [ ! -d debian/$(cdbs_curpkg)/usr/share/doc ] || \
    for dep in `perl -ne 'if (/^(Pre-)?Depends:/) {s/^\w+://; foreach (split 
/,/) { split; print($$_[0], "\n"); } }' debian/$(cdbs_curpkg)/DEBIAN/control`; 
do \
        if [ -d debian/$$dep/usr/share/doc ]; then \
                echo "Searching for duplicated docs in dependency $$dep..."; \
                rootdir=`pwd`; \
                (cd debian/$(cdbs_curpkg)/usr/share/doc/$(cdbs_curpkg); find 
-type f ! -name copyright | while read f; do \
                    
thisfile="$$rootdir/debian/$(cdbs_curpkg)/usr/share/doc/$(cdbs_curpkg)/$$f"; \
                    depfile="$$rootdir/debian/$$dep/usr/share/doc/$$dep/$$f"; \
                    if [ -f $$depfile -o -L $$depfile ] && zcmp $$thisfile 
$$depfile >/dev/null; then \
                        echo "  symlinking $$f in $(cdbs_curpkg) to file in 
$$dep"; \
                        rm $$thisfile; ln -s /usr/share/doc/$$dep/$$f 
$$thisfile; \
                    fi; \
                done ); \
            fi; \
    done

        # fix absolute symlinks created above to be relative
    dh_link -p $(cdbs_curpkg)

    dh_md5sums -p$(cdbs_curpkg) $(DEB_DH_MD5SUMS_ARGS)
    $(DEB_DH_BUILDDEB_ENV) dh_builddeb -p$(cdbs_curpkg) $(DEB_DH_BUILDDEB_ARGS)

Ubuntu ships a snippet in cdbs to fix this. Perhaps we could put this into haskell-devscripts if this is considered to be a problem. (attached)

signature.asc
Description: Digital signature