package changes size?

seth vidal skvidal at fedoraproject.org
Fri Oct 26 12:42:09 UTC 2007


On Fri, 2007-10-26 at 14:19 +0200, Thorsten Leemhuis wrote:
> On 26.10.2007 13:55, David Timms wrote:
> > David Boles wrote:
> >> on 10/26/2007 6:25 AM, David Timms wrote:
> >>> Peter Gordon wrote:
> >>>> On Thu, 2007-10-25 at 21:45 -0700, John Poelstra wrote:
> >>> And it would be nice for both of them to "talk" in the same language or 
> >>> better show both eg:
> >>> package      download size    install size
> >>> fred-1.0.1     100kB             665kB
> >>>
> >>> Along with the totals for the transaction set:
> >>>               4.678MB          13.434MB
> >> You are joking here correct? You 'expect' a compressed file to 'announce'
> >> it's installed size?
> > Yeah.
> > It wouldn't need to be perfect {eg when scripts generate further data 
> > during the install}.
> > What OS does that?
> > Like zipinfo. Maybe a new feature to include in rpm metadata itself ?
> > Maybe Fedora needs to be first modern one ?
> 
> It's likely easier -- the rpm header contains all the needed info
> already afaik. And it's even in the data that yum downloads afaics; take
> a look at the file /updates/7/x86_64/repodata/primary.xml.gz on your
> favorite mirror and you'll for each package see something like this:
> 
> [...]
>   <size package="400311" installed="1213843" archive="1218576"/>
> [...]
> 
> I suppose that's the data that is needed and it's likely in the
> primary.sqlite.bz2 (which yum prefers these days iirc) as well.
> 
> David, maybe just file a bug as RFE and see what happens.

I can tell you what'll happen :)

The issue is where to put the data on the screen and how to keep from
confusing the user. We could definitely say 'total installed size' in
the total summary but putting installed size next to download size for
each package just eats up valuable screen real-estate on a console.

So if you want to file a bug, no problem, just have a suggestion on
where you want this to show up.

-sv





More information about the fedora-test-list mailing list