* Upgrade to the latest stable version 6.2.8
* Move JS files to blender-cloud
* Introducing Hotkeys support (a'la YouTube)
* Introducing Loop button (and a way to easily add new buttons)
* Fix Analytics plugin to work with the VideoJS 6
* Minor style tweaks to work with the latest update
Exceptions aren't guaranteed to have a 'message' attribute. It does have
'args', but str(ex) is probably more useful as it's likely to include
the exception type.
This allows for wrapping the view function in another function which will provide additional args. Originally implemented in order to allow Dillo post view to provide the Project object to the view_embed template.
By specifying the 'embed_project' argument, the node's Project will be fetched and embedded in the document. This is useful in specific cases, where a project property needs to be accessed when rendering the view_embed template.
This 'custom view' means that the URL to view the node (as returned by
url_for_node(…)) cannot be loaded via XHR and embedded in the project
viewer, but should rather be used as the actual browser URL instead.
Currently only blogs use this.
- Include release stage, which should be 'production' or 'development',
and gets postfixed by '-debug' when running in debug mode.
- Properly logging remote IP address when proxied through HAProxy;
- Log user ID, email, username, roles, and capabilities;
- Remove authentication tokens from logged session;
- Log request data and JSON separately.
- Added request endpoint.
- Optimize JPEGs and increase quality from 75 to 95
- Don't always convert to RGB, first check if RGBA and save as PNG optimized
Thanks to Dr. Sybren and Francesco for review and feedback
There can be a KeyError accessing permission['methods'], but our current
logging doesn't provide enough information as to determine when this
happens. Rather than bluntly fixing the issue, I added logging to try and
find out how we get a 'methods'-less permission dict in the first place.
The response object *should* be a requests.Response object, which *should*
have a .text property. However, there are situations where this is not the
case, and in those cases we now won't produce an AttributeError.
This makes it easier to see what the Celery worker is actually working on
when refreshing a large number of links.
It'll report on every N refreshed links, where N = link_count/25 but
clamped to N ∈ [5, 100]
