It turns out this has to do with some additions to the fixture adding
of media entries. Since the new fake_upload field is True by default,
adding multiple entries that have media_files at once meant that
multiple additions of FileKeynames would be added at once if they were
not saved at the same time. Tricky!
Thankfully, Elrond helped us figure this one out. Thanks, Elrond :)
And thanks also to Aapo Rantalainen for sponsoring this commit!
A good idea, though it feels fairly clumsy in the form, and I think if
you're logged in you can already sabotage the user pretty well.
This commit sponsored by Sergey Matveev. Thanks!
- Added request.notifications
- Email configuration fixes
- Set config_spec default SMTP port to `0` and switch to SSL/non-SSL
default if `port == 0`
- Added email_smtp_use_ssl configuration setting
- Added migrations for notification tables
- Added __repr__ to MediaComment(Mixin)
- Added MediaComment.get_entry => MediaEntry
- Added CommentSubscription, CommentNotification, Notification,
ProcessingNotification tables
- Added notifications.task to celery init
- Fixed a bug in the video transcoder where pygst would hijack the
--help argument.
- Added notifications
- views
- silence
- subscribe
- routes
- utility methods
- celery task
- Added half-hearted .active comment CSS style
- Added quick JS to show header_dropdown
- Added fragment template to show notifications in header_dropdown
- Added fragment template to show subscribe/unsubscribe buttons on
media/comment pages
- Updated celery setup tests with notifications.task
- Tried to fix test_misc tests that I broke
- Added notification tests
- Added and extended tests.tools fixtures
- Integrated new notifications into media_home, media_post_comment views
- Bumped SQLAlchemy dependency to >= 0.8.0 since we need polymorphic for
the notifications to work
Tests:
- if the link doesn't exist yet
- if the link already exists
- if it is a symlink to something else
- if it exists but is not a symlink
This commit sponsored by Lon Koenig. Thank you!
It doesn't do much... it just has a static directory with a "bunnify"
CSS file and has a unit test that returns a json encoded set of a
staticdirect with both this plugin enabled and also
with (theoretically) a file distributed with MediaGoblin (it isn't
really, but in our case that doesn't matter.)
Also set up the static_setup hook / PluginStatic object for this
plugin; testing that with `./bin/gmg theme assetlink` seemed to work.
(And argh, need to update that command name, I know.)
This commit sponsored by Bradley Kuhn. Thanks bkuhn!
The test_user_dev (as opposed to user_dev) was a legacy before we had
each application running in its own directory (as they now do in
pytest). Move that name to just user_dev... this is more consistent
with the rest of our naming and will make writing these config files
easier. (If we want to test that changing these still works, that
should be a separate unit test with special config files.)
Additionally, add plugin static serving to the common test paste
config file.
This commit sponsored by Juan Jose Marin Martinez. Thank you!
This is possibly the cleaner approach to the problem solved
in the previous commit.
The previous commit makes sure, that testing is enabled at
import time. While this one enables it just before the
individual test is run. So this one takes a bit more CPU.
Maybe revert the previous commit?
References: 0536306048daa0970d2e43411ba2a9bf073e570e
This allows for modifying any context *right before render*, including
access to the variables that are passed in. This test takes advantage
of that and takes one of the variables, "doubleme", and modifies
it (doubles it!)
In our case it turns "happy" and "joy" into "happyhappy" and "joyjoy".
This commit sponsored by Mark Holmquist. Thank you!
This should remove a bunch of confusing cruft. I hate using that
environment variable! Also that old code was fragile.
This commit sponsored by Stephen Milton. Thanks!
Not the working solution, but getting there conceptually. Basically
we'll have a key with the view and the template as a tuple which is
the context hook that anyone can attach to.
However, some changes have still to be made:
- The unit test doesn't work yet and contains a set_trace ;)
- We'll probably switch the "view" component from being the callable
to the "urlgen"'able name per Elrond's suggestion
- Found a bug in unit tests related to running custom apps for
different configs... hm. I need to fix this!
Nonetheless, making progress.
This commit sponsored by... wait a minute... Christopher Webber?!
Loading the thumb/medium sizes from the config, saving
things to the db, and loading the image is now all done by
resize_tool. It still calls resize_image for the actual
work.
LazyProxy caches its value. So create a unit test for
lazy_pass_to_ugettext to make sure its returned proxy
actually gives a different translation when the locale
is switched.
The storage tests work in the system's tmpdir. The python
docs say, we should clean up after using things. Yes the
directory should be cleaned up on reboot, but if running
tests a lot, the tmpdir could fill up, so we should really
cleanup.
So use the new .delete_dir() on the storage interface to
cleanup test dirs and get them finally removed with
os.rmdir. All nicely packed into cleanup_storage().
