A branch for this has been released which covers everything we're going
to do for RTM, however we will want to revisit this post-RTM to work on
further savings.

Current memory savings have come from reducing the number of dictionary
and predictive text instances across threads and removing an individual
magnifier element from each key, replacing it with a single shared
component.

Most savings will need to come from the ubuntu-keyboard itself, rather
than from the maliit-framework. A quick analysis of the maliit-
framework's memory usage (~17-18mb with a dummy plugin loaded) shows
that the majority comes as a result of being a QGuiApplication (~13-14mb
on arm), which could only be addressed in Qt itself.

Future savings in the keyboard might be found by:

* Replacing hunspell with chromium's hunspell branch. This uses a binary
dictionary format that can be accessed directly from disk, rather than
being entirely loaded into memory. This is likely to save us a further
5-10MB.

* Investigating any other key elements that can be shared (e.g. one
giant shared MultiPointTouchArea), as we have a lot of keys any small
savings here add up quite quickly.

-- 
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/1350193

Title:
  maliit-server consuming large amounts of memory

To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu-keyboard/+bug/1350193/+subscriptions

-- 
ubuntu-bugs mailing list
[email protected]
https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs

Reply via email to