Re: How to make Packages file 50% smaller
It's easier than that, actually. Since the Packages files are updated on a
strictly daily basis in unstable (and even more rarely in testing and
stable), the server could keep a few diffs for n past updates to the
On the subject, I wouldn't mind people using something like xdelta for
upgrading .debs. Binary diff, mmmmm.
EFD1 AC6C 7ED5 E453 C367 AC7A B474 16E0 758D 7ED9
-----BEGIN GEEK CODE BLOCK-----
GCS/CM>CC/IT d- s:+ a16 C++(++++)>$ UL++++>$ P--- L++>++$ E+ W+(-) N+ o? K? w---()
!O !M !V PS+(++)>+ PE-(--) Y+>+ PGP t+>++ !5 X-- R>++ tv(+) b+(++) DI(+) D++
G>+++ e--> h! !r y>+++
------END GEEK CODE BLOCK------
On Sat, 3 Mar 2001, Kenneth Vestergaard Schmidt wrote:
> On Saturday 03 March 2001 01:41, Brian May wrote:
> > Does this file really *need* to contain long verbose descriptions of
> > packages?
> Okay, being a brand-new guy here, I might mistake something, but in my case,
> wouldn't this cut the long descriptions in e.g. aptitude? Because I wouldn't
> want that.
> On the other hand, I have about 4 mirrors for unstable, 3 mirrors for stable,
> and a few other deb-sources, and even on my 256kbit DSL it takes around 15
> minutes to do an 'apt-get update', which is pretty annoying, so a reduction
> in this time would be great. Possibly having only one repository for the long
> descriptions, and then downloading Packages-files without the long ones would
> be the solution, as you suggest.
> Another thing, which I have been wondering about - and please don't bash me
> too much for showing how little I knwo - is if you could download .diffs for
> the Packages files. Couldn't you supply a time-stamp, and the server would
> return a .diff? Or would that generate too much CPU-load?
> Anyway, that's my 2 cents :)
> To UNSUBSCRIBE, email to email@example.com
> with a subject of "unsubscribe". Trouble? Contact firstname.lastname@example.org