changelogs in packages and space use

[n0dalus]

On 8/31/07, seth vidal <skvidal at fedoraproject.org> wrote:
> Colin Walters mentioned on irc reducing the changelogs kept in pkgs a
> bit to help us fit on the livecd a bit better. I didn't really believe
> it would make enough of an impact to be noticeable but I did a little
> mucking about and found I was wrong. So I wrote this:
>
> http://skvidal.fedorapeople.org/changelog/changelog-sizes.py
>
> It'll dump out 4 files:
>
> - old-repo.txt : changelogs from the pkgs in the repos where the
> changelog entry is > 1yr old
> - new-repo.txt : changelogs from the pkgs in the repos where the
> changelog entry is < 1yr old
>
>
> - old-db.txt : changelogs from the pkgs in the rpmdb where the changelog
> entry is > 1yr old
> - new-db.txt : changelogs from the pkgs in the rpmdb where the changelog
> entry is < 1yr old
>
> On a mostly-rawhide system with rawhide repos I found:
>
> 8157 pkgs in rawhide repos:
>   new changelogs: 12M
>   old changelogs: 31M
>
> on my system 1128 packages installed:
>   new changelogs: 2.2M
>   old changelogs: 8.6M
>
>
> That may not seem like much but when we're squinched for space in 700M
> on a livecd that 8.6M might be all that's needed. Moreover it may be a
> bit silly to be carting all of this history around in all the pkgs and
> repodata and in the rpmdb. Esp if we're trying to trim download sizes
> for various items. So things we can do:
>
> 1. trim the changelogs at createrepo-runtime - fine - but that only gets
> it for the repodata
>
> 2. trim repos at rpmbuild time - great - I've suggested it as an option
> to rpmbuild on rpm-maint list.
>
> 3. trim them out of the pkgs the next time we change a package. Just
> prune them down to the last years worth of changelogs - maybe saving the
> old changelogs in a file in the cvs repository - or even into an unused
> source file in the srpm?
>
> What're people's thoughts on this?

If the changelog is trimmed, it would be nice if the full changelog
was provided in a new gzip compressed field. New versions of rpm would
read and see the full changelog, while old versions would just get the
trimmed version. On my system 8MB of changelogs generated by the
script compresses to 2MB.

Honestly though, I think there are probably better places to look for
space savings.

n0dalus.