* corrected merge-bug(?) in UtilController.php

* removed vendor from .gitignore
* updated and added dependencies in vendor folder

Conflicts:
	.gitignore
	README.md
	import/marc_local.properties
	solr/biblio/conf/schema.xml

- Bad interaction between Zend\Session\Container and proxy manager caused many notices.
- Came up with simpler mechanism for deferring Demo driver container instantiation.

Fixed fi and sv translation of Zip.

- Rewrote Javascript to make better use of server-side logic
- Eliminated many server-side special cases specific to lightbox handling
- Adjusted test suite to accommodate improvements

- Refactored use of Zend\Session\Container to inject proper session manager
- Introduced "on demand session container" mechanism for non-critical session data (only used by Demo ILS driver for now)
- Changed controllers to retrieve authorization service on demand rather than through injection (to defer session load)
- Replaced \VuFind\Controller\AbstractBase::writeSession() with \VuFind\Controller\AbstractBase::disableSessionWrites (and new related mechanism for safer AJAX session utilization)
- This makes sessions "on demand" which improves performance of requests that do not depend on session data (e.g. cover images, some AJAX calls).

corrects check for existence of ilslink

- Create generic (optional) caching mechanism in base class.
- Replace VoyagerRestful session-based caching with new generic mechanism.

- Thanks to @Cloud8 and @lahmann
- Resolves PR #612.

Claas Kazzer authored 9 years ago and André Lahmann committed 9 years ago

* fix for media queries, breakpoints

- This enables PDO to properly escape data sent to the bytea search_object field.
- Includes test scenario to prevent regressions.

- More specific check on return value for file_get_contents().

- We need to base DOM IDs on the FULL, UNIQUE identifier, not just the search class; otherwise, multiple filtered views of the same search backend will yield duplicate IDs.

- There was a race condition where multiple parallel requests could end up with a fatal error caused by one process retrieving the partially created row generated by another process and trying to unserialize an empty string. By waiting to write the session ID to the database until we also have the data available, we can avoid this problem.