> On Sept. 19, 2012, 1:58 p.m., Sebastian Trueg wrote: > > Do you have a more detailed callgrind analysis of the removable media > > model? Maybe we can optimize it after all. In the end it should only be > > some hash lookups and string replacement. That is nothing complicated. So > > if its slow the implementation might be bad.
It is just has lookups and string replacements, but they add up over time. When indexing files storeResources calls 1 query per statement + 2 query per resource (cardinality + types) + graph metadata checks + identification. On average it is very common to have 10-15 queries for even pushing very simple data (mimetype + stat result + rdf:types). Anyway, this patch seems fairly non-controversial. Doesn't it? - Vishesh ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: http://git.reviewboard.kde.org/r/106505/#review19162 ----------------------------------------------------------- On Sept. 19, 2012, 11:53 a.m., Vishesh Handa wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > http://git.reviewboard.kde.org/r/106505/ > ----------------------------------------------------------- > > (Updated Sept. 19, 2012, 11:53 a.m.) > > > Review request for Nepomuk and Sebastian Trueg. > > > Description > ------- > > When there are no removeable media connected we shouldn't have to pay > the extremely steep price of the RemoveableMediaModel. > > > Diffs > ----- > > common/removablemediacache.h ed11f7d > common/removablemediacache.cpp fcdec07 > services/storage/removablemediamodel.cpp d2be90a > > Diff: http://git.reviewboard.kde.org/r/106505/diff/ > > > Testing > ------- > > > Thanks, > > Vishesh Handa > >
_______________________________________________ Nepomuk mailing list [email protected] https://mail.kde.org/mailman/listinfo/nepomuk
