> On May 27, 2015, 3:05 p.m., Vishesh Handa wrote: > > src/file/extractor/app.cpp, line 140 > > <https://git.reviewboard.kde.org/r/123916/diff/1/?file=370952#file370952line140> > > > > Is the 'i:' needed? > > > > Same about the flushing.
Actually we could just output sucess or failiure(0/1 or anything), we already know what we're indexing, this is not final, I'll be changing it as per requirements later on. The flushing is needed, as we need to write as soon as possible so that we start with the next file. > On May 27, 2015, 3:05 p.m., Vishesh Handa wrote: > > src/file/fileindexingqueue.h, line 84 > > <https://git.reviewboard.kde.org/r/123916/diff/1/?file=370954#file370954line84> > > > > Given how complex the FileIndexingQueue is, and how it is a nightmare > > to test, in a future patch we should split out this process communication > > related code on both ends - here and in the extractor. Sounds like a good idea > On May 27, 2015, 3:05 p.m., Vishesh Handa wrote: > > src/file/extractor/app.cpp, line 110 > > <https://git.reviewboard.kde.org/r/123916/diff/1/?file=370952#file370952line110> > > > > You're now committing after each file. This is going to be terribly > > slow. > > > > But you can work on improving that in another patch. Re-running the indexing, i see it is excruciatingly slow, I'll have to work on batch committing fast. - Pinak ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://git.reviewboard.kde.org/r/123916/#review80902 ----------------------------------------------------------- On May 27, 2015, 12:37 p.m., Pinak Ahuja wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://git.reviewboard.kde.org/r/123916/ > ----------------------------------------------------------- > > (Updated May 27, 2015, 12:37 p.m.) > > > Review request for Baloo and Vishesh Handa. > > > Repository: baloo > > > Description > ------- > > With this patch the we start a single extractor process, and keep passing ids > to it one by one instead restarting it with 40 ids as arguments everytime. > This will make monitoring which file is being indexed easier. (might improve > performance too). This will also make implementing a monitor easier (which is > a part of my GSoC project). Right now we commit after every file which is not > ideal but this should be fixed in the next iteration. > > > Diffs > ----- > > autotests/unit/file/CMakeLists.txt fc0c70c > src/file/CMakeLists.txt 8158b7f > src/file/extractor/app.h 9a5112f > src/file/extractor/app.cpp 4fa28f0 > src/file/extractor/main.cpp beb6d60 > src/file/fileindexingqueue.h f17fb94 > src/file/fileindexingqueue.cpp 19ea53b > src/file/indexingqueue.h bc95c7c > > Diff: https://git.reviewboard.kde.org/r/123916/diff/ > > > Testing > ------- > > Baloo seems to be indexing files properly on my system without crashing. > > > Thanks, > > Pinak Ahuja > >
>> Visit http://mail.kde.org/mailman/listinfo/kde-devel#unsub to unsubscribe <<
