D12336: Replace several Q_ASSERTs with proper checks

Stefan BrĂ¼ns noreply at phabricator.kde.org
Sat Sep 29 16:14:24 BST 2018


bruns added inline comments.

INLINE COMMENTS

> apol wrote in documenturldb.cpp:44
> Would it make sense to include a warning then? If it's a wrong input, we'll need ways to debug it now that it won't be giving a backtrace.

This may be accessed quite frequently, so we would get a lot of warnings, spamming the journal, and slowing down the system.

A backtrace is useless here, as the code processes whatever it finds in the database. The interesting part is how the bad data ended up in the db.

Doing the checks here avoids spreading any existing corruption.

The long term solution is to:

1. Find the bugs leading to the DB corruption
2. Run lowlevel and highlevel sanity checks on existing databases

Both are on my TODO list, but this will take some time.

REPOSITORY
  R293 Baloo

REVISION DETAIL
  https://phabricator.kde.org/D12336

To: bruns, #baloo, michaelh, #frameworks
Cc: apol, sitter, kde-frameworks-devel, broulik, #frameworks, ashaposhnikov, michaelh, astippich, spoorun, ngraham, bruns, abrahams
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.kde.org/pipermail/kde-frameworks-devel/attachments/20180929/9550af23/attachment.html>


More information about the Kde-frameworks-devel mailing list