D16266: [Extractor] Make extractor crash resilient

[Stefan Brüns]

bruns marked 3 inline comments as done.
bruns added inline comments.

INLINE COMMENTS

> poboiko wrote in filecontentindexer.cpp:91
> There is `Transaction::documentId(const QByteArray& path)`, which can resolve it using `DocumentUrlDB`.
> I believe it still would be cheaper than reindexing several files multiple times.

But it is racy - if the file is replaced in the meantime, inode and filename no longer match. This is not completely unlikely when dealing with temporary files.

It would make the code also significantly more complex, and I want it simple here. It should only be hit in exceptional cases.

Also, it is not to uncommon to have a batch of size one from the start, i.e. when adding/modifying files.

REPOSITORY
  R293 Baloo

REVISION DETAIL
  https://phabricator.kde.org/D16266

To: bruns, #baloo, #frameworks, poboiko, ngraham
Cc: broulik, apol, kde-frameworks-devel, ashaposhnikov, michaelh, astippich, spoorun, ngraham, bruns, abrahams
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.kde.org/pipermail/kde-frameworks-devel/attachments/20181017/bfd6e219/attachment-0001.html>