Hi Martin! I will tell you a dirty secret:
Nokia-N900:~# ls -l /usr/share/ | grep icons lrwxrwxrwx 1 root root 20 Jul 13 14:27 icons -> /opt/usr/share/icons Thanks, Daniil. On Wed, Jul 21, 2010 at 4:50 PM, Marius Vollmer <marius.voll...@nokia.com> wrote: > ext Martin Storsjö <mar...@martin.st> writes: > >> OPT folder value should be 80% of USR folder. > > Taken literally, this means that the OPT folder is actually supposed to > be smaller than the USR folder... :-) > >> ACTUAL OUTCOME: >> Opt folder value is not 80% of usr folder ,usr=84,opt=388k. > > This actually checks out for me: > > 388 / (388 + 84) = 0.82 > 0.8 > > Maybe they want USR to be at most 20% of OPT? That would be a bit strange, > but > would explain the verdict: > > 84 / 388 = 0.21 > 0.20 > > If I remember right, Urho and me invented these numbers hand wavingly > some time ago. The idea back then was that if the OPT/USR ratio of a > package is always larger than the OPT/USR ration of the actual > partitions, then you can install as many packages as will fit into OPT > without running out of USR. > > The partitions are 2G / 256M = 8. > > With the "du" numbers on ext3, your package has a ratio of > > 388 / 84 = 4.6 < 8 > > which would indeed fail the requirements. > > > With your numbers on ubifs, we have > > 368 / 20 = 18.4 > 8 > > and you win by a huge margin. > > However, I agree that directories should not be counted at all. Then > you will probably win by a even larger margin. > > > To summarize: I don't know what they have calculated, but I think they > should calculate the OPT/USR ratio and require it to be larger than 8. > Furthermore, only actual files should be counted. Filesystem specific > effects like directories should be exluded. This is a heuristic anyway > and making it simpler and easier to reproduce in different environments > is therefore more important than accurate simulation of what would > happen on the device. > > (If we want to simulate the device, we need to take UBIFS compression > into account.) > >> Also, is there any idea in trying to discuss these issues in Ovi Publish >> by commenting back? On Ovi/Symbian things, I've never ever seen a reviewer >> actually take my comments in account, so I don't think anybody ever reads >> the comments the submitter writes. > > I'll try to ping the right people here. > _______________________________________________ > maemo-developers mailing list > maemo-developers@maemo.org > https://lists.maemo.org/mailman/listinfo/maemo-developers > _______________________________________________ maemo-developers mailing list maemo-developers@maemo.org https://lists.maemo.org/mailman/listinfo/maemo-developers
