* Add ctools, prior to using it.
* Fix up all the dependency references.
... before the colon is the project name, so should only be "drupal" for
modules shipped in core.
* Some more together.
* Decent progress... getting things actually rendering...
... bit of refactoring stuff making a mess.
* More worky.
... as in, basically functional. Still needs coding standards pass, and
testing with more/all types of content.
* Coding standards, and warning of validation issues.
* Pull the batch out to a separate service.
* Something of namespacing the child-specific batch...
... 'cause need to slap together a media-specific batch similarly?
* All together, I think...
Both the child-uploading, and media-uploading forms.
* It is not necessary to explicitly mark the files as permanent.
* Further generalizing...
... no longer necessarily trying to load files, where files might not
be present (for non-file media... oEmbed things?).
* Adjust class comment.
* Get rid of the deprecation flags.
* Remove unused constant.
... is defined instead at the "FileSelectionForm" level, accidentally
left it here from intermediate implementation state.
* Pass the renderer along, with the version constraint.
* Add update hook to enable ctools in sites where it may not be.
... as it's now required.
* Cover ALL the exits.
* Refine message.
* Excessively long line in comment...
... whoops.
* Bump spec up to allow ctools 4.
Gave it a run through here, and seemed to work fine; however, ctools'
project page still seems to suggest the 3 major version should be
preferred... but let's allow 4, if people are using or want to test it
out?
* Fix undefined "count" index.
* Add ancestor condition.
* Move the ancestors to the utils helper, re-work the builder.
* Use the config option as opposed to hard coding.
* Handle a looping scenario (a > b > c > a).
* More flexible headers for STOMP messages, theoretically.
* Lower the priority of the base implementation.
* More context.
* Add in the "aud" claim.
... ensure we're dealing with our tokens.
* Rectify the other base class.
* Remove some chaff intro'd from merge conflicts.
* Return after failing to find the "aud" claim.
* Permissively allow without the "aud" claim...
... _could_ roll more conditionally, with some state set during an update
hook; however, seems like unnecessary complexity.
* Couple of coding standards things.
* Add the use of the class back in.
Jared Whiklo
Alexander O'Neill