Remove use of final deprecated functions. Sadly these tests were not
properly updated when the functions were deprecated and continued to use
the deprecated functions. I've removed the uses of the final deprecated
functions and updated a few of the non-final-deprecation uses (but not
all)
This change removes the complex and partial loading of role definition data for users.
It is replaced by keeping one system-wide definition for each role in MUC.
This is cheaper to regenerate as it avoids complex database queries (repeated LEFT JOIN to context).
Memory usage is kept low thanks to array CoW for each user access data.
This version includes a fix[1] contributed by Frédéric Massart investigating
our compilation time issues, which significantly improves the speed of
processing font awesome.
Thanks, Fred!
[1] https://github.com/sabberworm/PHP-CSS-Parser/pull/120
Previously there were a few issues with the code
* We were capturing a promise without then then .then() that came after
it, so our promise wouldn't always be completely resolved by the time
then next piece of code was operating on it
* We weren't catching all errors with .catch()
These new methods will be used to render the tree of file types groups
and extensions in the browser widget.
Co-Authored-By: John Okely <john@moodle.com>
This is a stripped down version of the original Jonathon's element
without the actual JS selector. The options "allowall" and "onlytypes"
are not fully supported right now yet.
Starting with npm version 5, npm install will generate a lockfile
(now named package-lock.json) or update the existing npm-shrinkwrap.json
To prevent changes being generated in the codebase when this happens, we
need to commit the version 5 generated shrinkwrap.
This is related to a general overhaul [1] of how npm manages things. But
for our purposes, we the lockfile should be BC to previous versions [2].
Going forward, we need to ensure we generate the lockfile on npm >=5 to
prevent differences from the previously un-versioned lockfile.
[1] http://blog.npmjs.org/post/161081169345/v500
[2] https://github.com/npm/npm/issues/16728#issuecomment-305104149
Allow Global Search 3rd party engines to be much faster by supporting batch processing.
Refactor the iterator loop for the documents to add to the search index into its own method.
Move this new method from the manger class to the base engine class.
The move to the base engine class will allow search engine plugins to override this and determine
how they implment it. It will not break the existing interface contract with existing plugins.
Finally, add an elasped time indication to the trace output.
