hi
I recollected my earlier experiments on computing package diffs, and
prepared a package called 'debdelta'. It is available at
http://tonelli.sns.it/pub/mennucc1/debdelta and also uploaded into
experimental.
This package contains both a command 'debdelta', to compute deltas of
Debian packag
hi
I had the same idea some time ago
if you ever decide to work on that, I may help
Goswin von Brederlow wrote:
> I actualy have a little hack how one could implement patch debs now to
> test this out:
>
> 1. Create an archive mirror with rsync batch files (or xdelta or
> whatever) between the
Marc Haber <[EMAIL PROTECTED]> writes:
> On Mon, 01 May 2006 09:30:55 +0200, Florian Weimer <[EMAIL PROTECTED]>
> wrote:
>>The downside is that anything that doesn't work on entire .debs is
>>very likely to change them at the byte stream level (you only need to
>>use slightly different zlib versio
On Mon, 01 May 2006 09:30:55 +0200, Florian Weimer <[EMAIL PROTECTED]>
wrote:
>The downside is that anything that doesn't work on entire .debs is
>very likely to change them at the byte stream level (you only need to
>use slightly different zlib versions or parameters). This means that
>the chain
* Darren Salt ([EMAIL PROTECTED]) [060502 19:03]:
> I demand that Andreas Barth may or may not have written...
> > * Brian Eaton ([EMAIL PROTECTED]) [060501 19:21]:
> >> On 5/1/06, Andreas Barth <[EMAIL PROTECTED]> wrote:
> >>> Or you could create the diffdebs before upload or on ftp-master, and
>
I demand that Andreas Barth may or may not have written...
> * Brian Eaton ([EMAIL PROTECTED]) [060501 19:21]:
>> On 5/1/06, Andreas Barth <[EMAIL PROTECTED]> wrote:
>>> Or you could create the diffdebs before upload or on ftp-master, and
>>> include the diffdebs somehow in the Packages file (so t
I demand that Pierre Habouzit may or may not have written...
[snip; delta packages?]
> The real question is: do people clean their apt cache or not? I do, because
> after a full X.org/kde/openoffice upgrade, it takes quite a lot of disk in
> /var (that is small on my computers). And with that cach
Pierre Habouzit <[EMAIL PROTECTED]> writes:
> Le Lun 1 Mai 2006 15:31, Brian Eaton a écrit :
>> On 4/30/06, Goswin von Brederlow wrote:
>> > Look at zsync and help develope it far enough so it can look into
>> > debs. Without that the gain is practicaly 0 or less.
>>
>> It's entirely possible tha
Peter Samuelson <[EMAIL PROTECTED]> writes:
>> * Goswin von Brederlow:
>> > Look at zsync and help develope it far enough so it can look into
>> > debs. Without that the gain is practicaly 0 or less.
>
> The other thing to do would be to lobby for dpkg-deb and dpkg-source to
> use 'gzip --rsyncabl
Tyler MacDonald <[EMAIL PROTECTED]> writes:
> Goswin von Brederlow <[EMAIL PROTECTED]> wrote:
>> Bittorrent has a per chunk hash so it can validate each chunk when it
>> recieves it instead of waiting for the full file. It won't see if a
>> chunk is present at some other position in the file, not
hi
I did a similar thing some time ago; I used 'xdelta' on two versions of
kernel and of tetex; the results were impressive; I could prepare a
'debdiff' that was < 10% (AFAICR) of the size, and that would recreate
an exact copy of the new version of the package, given the previous
version of the p
* Brian Eaton ([EMAIL PROTECTED]) [060501 19:21]:
> On 5/1/06, Andreas Barth <[EMAIL PROTECTED]> wrote:
> >Or you could create the diffdebs before upload or on ftp-master, and
> >include the diffdebs somehow in the Packages file (so they're signed as
> >well by the usual mechanismn).
> My initial
On 5/1/06, Andreas Barth <[EMAIL PROTECTED]> wrote:
Or you could create the diffdebs before upload or on ftp-master, and
include the diffdebs somehow in the Packages file (so they're signed as
well by the usual mechanismn).
My initial view is that any delta package system that doesn't
reproduce
On 5/1/06, Andreas Barth <[EMAIL PROTECTED]> wrote:
* Brian Eaton ([EMAIL PROTECTED]) [060501 16:42]:
> On 5/1/06, Andreas Barth <[EMAIL PROTECTED]> wrote:
> >
> >If one does it right, it might be enough if the original package is
> >*installed*. And that happens quite often, e.g. even for securi
* Brian Eaton ([EMAIL PROTECTED]) [060501 17:49]:
> On 5/1/06, Andreas Barth <[EMAIL PROTECTED]> wrote:
> >* Brian Eaton ([EMAIL PROTECTED]) [060501 16:42]:
> >> On 5/1/06, Andreas Barth <[EMAIL PROTECTED]> wrote:
> >> >
> >> >If one does it right, it might be enough if the original package is
> >>
* Brian Eaton ([EMAIL PROTECTED]) [060501 16:42]:
> On 5/1/06, Andreas Barth <[EMAIL PROTECTED]> wrote:
> >* Brian Eaton ([EMAIL PROTECTED]) [060501 15:51]:
> >> The only time delta packages will be a win is for upgrades where the
> >> client has the original package cached.
> >
> >If one does it r
Le Lun 1 Mai 2006 16:35, Brian Eaton a écrit :
> On 5/1/06, Andreas Barth <[EMAIL PROTECTED]> wrote:
> > * Brian Eaton ([EMAIL PROTECTED]) [060501 15:51]:
> > > The only time delta packages will be a win is for upgrades where
> > > the client has the original package cached.
> >
> > If one does it
On 5/1/06, Andreas Barth <[EMAIL PROTECTED]> wrote:
* Brian Eaton ([EMAIL PROTECTED]) [060501 15:51]:
> The only time delta packages will be a win is for upgrades where the
> client has the original package cached.
If one does it right, it might be enough if the original package is
*installed*.
Le Lun 1 Mai 2006 15:31, Brian Eaton a écrit :
> On 4/30/06, Goswin von Brederlow wrote:
> > Look at zsync and help develope it far enough so it can look into
> > debs. Without that the gain is practicaly 0 or less.
>
> It's entirely possible that the gain will be nothing no matter what
> algorithm
* Brian Eaton ([EMAIL PROTECTED]) [060501 15:51]:
> The only time delta packages will be a win is for upgrades where the
> client has the original package cached.
If one does it right, it might be enough if the original package is
*installed*. And that happens quite often, e.g. even for security
On 4/30/06, Goswin von Brederlow <[EMAIL PROTECTED]> wrote:
> Brian Eaton <[EMAIL PROTECTED]> wrote:
>> http://rsync.samba.org/rsync-and-debian/rsync-and-debian.html
>>
>> Has anyone ever done some log file analysis to figure out how much
>> bandwidth would be saved by transferring package deltas
On 5/1/06, Brian Eaton <[EMAIL PROTECTED]> wrote:
Hello all -
Regarding the ideas discussed here:
http://rsync.samba.org/rsync-and-debian/rsync-and-debian.html
A few comments:
3.2 rsync is too hard on servers
That document claims it's an avoidable cost. It's not really, because
the client p
> * Goswin von Brederlow:
> > Look at zsync and help develope it far enough so it can look into
> > debs. Without that the gain is practicaly 0 or less.
The other thing to do would be to lobby for dpkg-deb and dpkg-source to
use 'gzip --rsyncable' when building stuff. (That, or sneak
"--rsyncabl
* Goswin von Brederlow:
> Tyler MacDonald <[EMAIL PROTECTED]> writes:
>
>> Brian Eaton <[EMAIL PROTECTED]> wrote:
>>> http://rsync.samba.org/rsync-and-debian/rsync-and-debian.html
>>>
>>> Has anyone ever done some log file analysis to figure out how much
>>> bandwidth would be saved by transferri
Goswin von Brederlow <[EMAIL PROTECTED]> wrote:
> Bittorrent has a per chunk hash so it can validate each chunk when it
> recieves it instead of waiting for the full file. It won't see if a
> chunk is present at some other position in the file, not even if that
> position is also on chunk boundarie
Tyler MacDonald <[EMAIL PROTECTED]> writes:
> Brian Eaton <[EMAIL PROTECTED]> wrote:
>> http://rsync.samba.org/rsync-and-debian/rsync-and-debian.html
>>
>> Has anyone ever done some log file analysis to figure out how much
>> bandwidth would be saved by transferring package deltas instead of
>> e
Brian Eaton <[EMAIL PROTECTED]> wrote:
> http://rsync.samba.org/rsync-and-debian/rsync-and-debian.html
>
> Has anyone ever done some log file analysis to figure out how much
> bandwidth would be saved by transferring package deltas instead of
> entire new packages?
Slightly off-topic, but
Hello all -
Regarding the ideas discussed here:
http://rsync.samba.org/rsync-and-debian/rsync-and-debian.html
Has anyone ever done some log file analysis to figure out how much
bandwidth would be saved by transferring package deltas instead of
entire new packages?
Assuming someone hasn't done
28 matches
Mail list logo
