Duplicated files in the pristine FC4t2 installation

Hans de Goede j.w.r.degoede at hhs.nl
Mon May 2 12:52:02 UTC 2005


Hmm,

Lots of pyo pyc duplicates, this should be somehow fixed in python can 
RPM handle hardlinks iow can an rpm contain a file and hardlink to the 
file instead of 2 copies of the file?

If rpm can handle hardlinks then this should be fixable preferrably 
python should just create a hardlink when the pyc and pyo are the same.

About licenses what about a licenses.rpm which gets installed by default 
which contains most common licenses and which is an obligatory part of 
the base system. And then just put a ptr to the license in the rpm -qi 
info? That would also cleanout /usr/share/doc since quite a few dirs 
there only contain a copy of the GPL / MPL / whatever license

Regards,

Hans





Jindrich Novy wrote:
> Hello all,
> 
> I've found some file duplicates when I browsed through the /usr
> directory tree in the pristine & complete FC4t2 installation what made
> me curious how many duplicates are there in total. This is not critical,
> so please take this as something for your information that some files
> would better be symlinked/hardlinked in order to not to waste disc space
> without a point. I know there's sometimes no other way that to duplicate
> a file, but the statistics I have is IMHO rather interesting:
> 
> 206405 regular files found, 4149 MiB [5468 MiB]
> 15797 total dupes, 15705 non-zero sized.
> 96 MiB [161 MiB], 2.325% [2.951%] wasted by dupes, 13906 symlinks, 5042
> hardlinks.
> 
> So that 161 MiB is "wasted" physically in the /usr tree, what is about
> 3% in total from all the files within the /usr hierarchy.
> 
> To let this information be somehow worth for the package maintainers,
> I'm adding a link to the list of all the duplicated files including
> their sizes and md5 sums and to what package they belong:
> 
> http://people.redhat.com/jnovy/files/FC4t2-usr-dupes.gz
> 
> This statistics was done by the "slink" utility I wrote some time ago.
> It's able to replace duplicates with symbolic links to save disc space
> [EXPERIMENTAL, but seems to work] or just display a statistics about
> duplicates for a given directory. If you want to give it a try, get it
> from:
> 
> http://people.redhat.com/jnovy/files/slink-0.0.1-pre1.tar.bz2
> 
> It's interesting how many GPL "COPYING" clones we have in /usr/share/doc
> for instance. Unfortunately some of my packages are also affected ;)
> 
> Regards,
> Jindrich
> 





More information about the devel mailing list