D7995: KDevelop: address dirwatching inefficiency (WIP/PoC)

[Milian Wolff]

mwolff added a comment.


  In https://phabricator.kde.org/D7995#171218, @rjvbb wrote:
  
  > I found and will upload another solution which gets rid of the whole signal/slot overhead and potential risk of delivering signals to a stale object. Hand off a pointer to the watcher instance to the FIleManagerListJob instance and do the addDir() call directly from there. 
  >  Should maybe use a QSharedPointer or similar wrapper but that's a later concern.
  >
  > Is the dirwatching feature
  >
  > > still, doesn't change the fact that the whole watcher wrapper should be removed
  >
  > No. Reasons given before that I'm not going to repeat - it should be even clearer now that KDW::addDir() shouldn't be called directly each time a directory is (re)loaded.
  
  
  It's not being called directly, it's being called indirectly via the eventloop which is fine as long as you setup the connection properly, which you didn't do here.
  
  > There's another concern however: 
  >  https://bugs.kde.org/show_bug.cgi?id=387238

REPOSITORY
  R32 KDevelop

REVISION DETAIL
  https://phabricator.kde.org/D7995

To: rjvbb, #kdevelop, mwolff
Cc: aaronpuchert, arrowdodger, kfunk, dfaure, mwolff, brauch, kdevelop-devel, njensen, geetamc, Pilzschaf, akshaydeo, surgenight
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.kde.org/pipermail/kdevelop-devel/attachments/20171123/3226c15f/attachment.html>