What I currently have implemented is something like: If newRpmSize < 1M ==> drpm is not worth it if drpmsize > 50% of new rpm size ==> drpm is not worth it The actual values can be tuned of course, but I guess the idea of a lower bound, coupled with a percentage saving is generally acceptable. Also, preliminary size tests shows big savings for huge packages, so we only need to consider smaller ones carefully. <div>On 3/11/07, Warren Togami <<a href="mailto:wtogami@redhat.com">wtogami@redhat.com</a>> wrote:<blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;"> Toshio Kuratomi wrote: >> If you had efficiency threashold at >50% and size of ORIGINAL rpm >> larger than a certain size (say 1 meg? 10 megs? that would be up to >> whoever is doing the work and not me, a lurker), it could work. >> > Another metric could simply be absolute size of the difference. If a > major openoffice.org-core update was only a 25% savings, that would > still be about 25MB of data. > Err, yes, this is what I meant. =) Warren Togami <a href="mailto:wtogami@redhat.com">wtogami@redhat.com</a> -- fedora-devel-list mailing list <a href="mailto:fedora-devel-list@redhat.com">fedora-devel-list@redhat.com</a> <a href="https://www.redhat.com/mailman/listinfo/fedora-devel-list"> https://www.redhat.com/mailman/listinfo/fedora-devel-list</a> </blockquote></div>