aacid added a comment.
In https://phabricator.kde.org/D7671#142844, @rkflx wrote: > Albert: Do you remember why https://phabricator.kde.org/R223:f93ccd7923491c6b1412ba5cb1fe0711e44496d8 was necessary? There is no bug linked and no autotest. Using just the commit message as a testcase, with https://phabricator.kde.org/D7671 applied and `watcher->addDir(fi.absolutePath())` removed I can still trigger a reload with `rm file.txt; touch file.txt`. At first sight, watching a directory when we are interested in the file feels like a workaround for bugs or missing features in KDirWatch at the time (and not needed anymore today), perhaps something like inode vs. path watchers. > > We could think about whether https://phabricator.kde.org/R223:f93ccd7923491c6b1412ba5cb1fe0711e44496d8 can be reverted in total (not sure, did not grasp the other code parts yet) in favour of https://phabricator.kde.org/D7671. The former seems complicated and does not support Kate, the latter is easy to understand and supports both the original case and Kate. > > If there is interest, I can provide logs from the ancient system gathered with `inotifywait -m .` which show what's going on. > > (Julian: Sorry to intrude, feel free to pick up from here.) It was added to make this work, and it did work at some point, i don't add code for nothing ;) REPOSITORY R223 Okular REVISION DETAIL https://phabricator.kde.org/D7671 To: progwolff, aacid Cc: sander, rkflx, #okular, aacid
