> On Feb. 4, 2017, 7:59 p.m., Albert Astals Cid wrote: > > Any reason this has not been commited? > > Luigi Toscano wrote: > Because there was no final decision. > > It is true that the Baloo database can be regenerated, but I would say > that if we avoid a complete rebuild of the database after a Frameworks > update, that would be good. > > Would it be better to read the old location if found and reserve the new > one only for new database (or rebuilt) and switch to the new one for > Frameworks 6? > > James Smith wrote: > Regeneratable data isn't config, so this change isn't as important as the > config location change between SC4 and KF5. The config data location switch > was mandatory for KF5-based apps. I think we can safely switch regeneratable > data at any time. > > Luigi Toscano wrote: > Did you try it in practice? I'm talking about time and I/O which is not > visibile in the normal usage because the full indexing already happened. > Moreover the old index will be left there unused. > > James Smith wrote: > A 64069 file reindex for me took ~58 mins. An update of eg. Calligra > shouldn't be done when there is an immediate requirement to use Calligra, > likewise an update of Baloo shouldn't be attempted when there is expected use > of Baloo in the immediate future; don't fix it if it isn't broken. There has > been a suggestion to group patches that require a reindex, thereby lessening > any reindexing caused by upgrading Baloo. > > The old index would be left there unused whether the patch is waited on > for KF5's successor or otherwise. > > Luigi Toscano wrote: > It does not work like that. Distributions like Fedora, Archlinux, or > Gentoo, and also Debian unstable, upgrades the Frameworks packages > frequently, so their users will be impacted. > > It is more expected to have leftovers with a major upgrade, but > nevertheless it would be good to have a migration. > > So to summarize I think that: > - the old location should be used if available, or > - if a migration should be forced, the data should be moved (in an async > way - is there anything that prevent that? Could it happen that they are on > different drives?)
As a related fact, I have two pending review requests (#129839 and #129798) on KFileMetaData that may require reindexing audio files. Not sure there is a good way to do that but definitely related to Baloo migration issues. - Matthieu ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://git.reviewboard.kde.org/r/129605/#review102382 ----------------------------------------------------------- On Dec. 12, 2016, 12:06 p.m., James Smith wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://git.reviewboard.kde.org/r/129605/ > ----------------------------------------------------------- > > (Updated Dec. 12, 2016, 12:06 p.m.) > > > Review request for Baloo and Vishesh Handa. > > > Repository: baloo > > > Description > ------- > > Move the database into the XDG cache path. > > > Diffs > ----- > > src/engine/global.cpp 4f511ac08f0fd18bd3c78b794f4ba7ccef88e75b > src/tools/balooctl/main.cpp 2a6b175ee599d7ca713adcbed80d3bd128e9dd63 > > Diff: https://git.reviewboard.kde.org/r/129605/diff/ > > > Testing > ------- > > Compile, run. > > > Thanks, > > James Smith > >
