Robert Collins wrote: [snip] > > > Then the client can trivially choose the most bw efficient way to > > > receive the update. > > > > No need for that one, if deltas larger than the original are simply not > > created on the server. AFAICS the index file should look like this: > > I presume you mean deltas with an aggregate size > than the current > Packages ? Exactly. > > 20040703001 Packages.gz > > 20040702001 Packages-cdiff-20040702001.gz > > 20040701001 Packages-cdiff-20040701001.gz > > 20040630001 Packages-cdiff-20040630001.gz > > 20040629001 Packages-cdiff-20040629001.gz > > ... > > > > with an cutoff based on cdiff size and age. That's some redundancy, > > but it simplifies processing. And apt-get doesn't have to make > > assumptions about the server's update frequency, clocks, timezones, > > etc. > > I don't think that adding the file size forces apt-get to make any > assumptions about update frequency. This was a general remark about the format described above. :-) > But it does allow more client side > optimisations in the future. (There are more factors than raw size to > consider - round trips, processing overhead etc. Er, which one exactly? For round trip time, we should keep the size of the index file small, because it has to be fetched instead of the Packages file timestamp, even for unsuccessful (already-updated) attempts. > Giving the client the > info to optimise as needed in the future makes sense to me) But we optimize for low server load. The client load is mostly irrelevant. Thiemo
Attachment:
signature.asc
Description: Digital signature