Attending: Guenter Hipler, Demian Katz, Claas Kazzer, André Lahmann, Jochen Lienhard, Ere Maijala, Brad Patton

• VUFIND-1131 - This ticket suggests a UI improvement for renewals.
• VUFIND-1133 - This ticket relates to the fact that auto-focus of search boxes is not currently working due to a bug in the Bootstrap autocomplete library.
• VUFIND-1134 - This ticket requests a Libraryh3lp chat plugin; Demian has implemented one.

Several spam tickets have been deleted.

• #496 - merged - improved configuration reading for combined search.
• #497 - in progress - work on relators for the “improved author indexing” PR.
• #498 - merged - bug fix for retrieving multiple records by ID while sharding.
• #499 - merged - made the “similar” recommendation module backend-agnostic (was previously hard-coded to “Solr”).
• #500 - in progress - work on adding shard configuration to combined search columns.
• #501 - in progress - work on removing bundled dependencies from VuFind.
• #502, #503 - merged - small fixes/improvements to AJAX tab loading.
• #504 - merged - Translation fix.
• #505 - merged - Template typo fix.
• #506 - merged - Adds comment to httpd-vufind.conf regarding configuration of Shibboleth with Apache 2.4 (previous comment was 2.2-specific).

André and Demian have been working on getting the indexing and display of author relators right; a solution has nearly been reached. Some progress has also been made on indexing corporate authors separately from non-corporate authors to reduce the amount of MARC-specific logic coded into VuFind's PHP.

Demian has been doing some research into docValues. See comments on VUFIND-1100 for more details. He will continue experimenting in the near future.

More tests have been written, and the test suite has been adjusted to allow both Selenium and Zombie.js drivers to be used (for either convenient headless testing, or for the ability to watch the tests executing in a real browser). Unfortunately, there are still some challenges getting the tests to pass consistently – nothing wrong with VuFind itself but rather with the inherent challenges of web testing. Chris and Demian continue to work on this.

Two main topics from Guenter:

1.) Refactor application to be more component-based.

2.) Try to build API for some functionality.

These goals should be approached incrementally. First step: spend two or three months gathering information, getting an idea of what might be possible.

One example: factoring out covers to a separate module.

Another: for next year, Markus making steps in the direction of an API.

Guenter has built a search API prototype using APIgility. Ere is also working on an API prototype (see here – but note that this is in early stages; handle with care).

Demian favored initially searching for “low hanging fruit” to factor out to make the core smaller and easier to analyze the more complex problems.

André pointed out that components may help spread out maintenance responsibility for the project.

These items were noted down at the VuFind Summit; we should determine which should move to the ongoing development discussion, which should become JIRA tickets, who should work on what, etc.

Refactoring of Modules

• Separate modules for search backends?
• How much is too much?
• ILS drivers to a module with DAIA/PAIA interface

API

• Where? How?
• Simple extension of existing RSS feed? What else?
• Consider on a per-module basis for organization
• Needs to be extendable

Permissions

• Move forward with Demian's email proposal

Language Namespacing

• Tokenize more, chunk less
• API strings are a problem (EDS)
• Move towards Translate Wiki

Minimum system requirements

• Solr 5?
• PHP 5.5?
• Compare to requirements of Zend Framework 3

Should we continue to bundle the vendor directory?

• Leaning towards no
• Install package would be fine
• Git hooks

More asynchronous loading of inline elements

• Holdings Tab
• More Like This

Investigate new Solr spelling options

Support Solr date range fields (Solr 5)

Revisit binary availability status in ILS interface.

Cover generator options

• Cache proxied urls
• Pattern-based image retrieval from local file system

Better documentation about command line tools, LESS, etc.

There has been some renewed discussion about the call platform; there may be a need to find a more Linux-friendly call mechanism without the 10-person limitation of Google Hangouts.

Due to the Thanksgiving holiday, November 24 is not going to be a good date for a call. Instead, we will wait for December 8.

The next call will be Tuesday, December 8, 2015 at 10am Eastern Standard Time (15:00 GMT).