dfaure added inline comments.
INLINE COMMENTS
> kdirwatch.cpp:721
> QFile::encodeName(e->path).data(), mask))
> >= 0) {
> + m_inotify_wd_to_entry[e->wd] = e;
> if (s_verboseDebug) {
.insert(e->wd, e) is a hair faster, says Effective STL Item 24 :)
> rjvbb wrote in kdirwatch_p.h:224
> How much overhead does this give per watched item?
>
> I mean in bytes but I'm guessing there might be a break-even point, a number
> of entries under which the look-up in the map is more expensive than the
> current implementation. That could affect applications using KDW to monitor a
> tiny number of files. If correct, what kind of overhead are we talking about?
Each hash node will be approx 4+8=12 bytes, and something must point to it, so
another 8 bytes, 20 in total. No big deal.
REPOSITORY
R244 KCoreAddons
REVISION DETAIL
https://phabricator.kde.org/D9824
To: mwolff, dfaure, rjvbb, #kdevelop
Cc: #frameworks
