http://bugzilla.moblin.org/show_bug.cgi?id=2229
pohly <[email protected]> changed: What |Removed |Added ---------------------------------------------------------------------------- AssignedTo|[email protected] |[email protected] --- Comment #12 from pohly <[email protected]> 2009-09-21 03:09:15 --- (In reply to comment #11) > (In reply to comment #10) > > > But the prepares are almost NOPs. There isn't much to do when checking the > > items, so this part is done with very quickly. If the weights are meant to > > turn > > item counts into real time, then 0.1/0.45/0.45 might be more suitable. > > Not item counts, I was trying to estimate the total time in a common sync > situation, which at least for me meant hundreds of prepares and 0-3 sends and > receives. > > In the normal case 0.1/0.45/0.45 means smooth progress (with some values of > smooth) until ~15% and then basically jumping to complete. Currently this > means > smooth and fast progress until ~50% and then jumping to complete. So in other words, giving more weight to the prepare phase is better because the actual data exchange is quick. > The other end of the spectrum is of course the first sync with all the data on > one end (let's say server): > 0.1/0.45/0.45 would mean that the progress jumps to ~60% and then slowly fills > up. > Current ratios mean progress jumps to ~75% and then very slowly fills up. I suppose the bottom line is that fixed ratios between the phases don't work. They need to be dynamic, based on whether we expect to really transfer items. This is tricky to implement. We might have a better chance to implement it in the D-Bus server. > It is possible that the current 'bias' in the weights is a result of using > very > large test databases so prepare seemed to take a longer time. I'll be happy to > change if you want to. No, don't bother. I'm taking this issue. I need to implement the "waiting/processing" distinction in the new D-Bus API (for the spinner), then look into better estimating the progress. -- Configure bugmail: http://bugzilla.moblin.org/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are watching someone on the CC list of the bug. _______________________________________________ Syncevolution-issues mailing list [email protected] http://lists.syncevolution.org/listinfo/syncevolution-issues
