Tags:

I just posted a new build of TagLib. It has a bunch of bugfixes, a little new feature, more stable code, and one major flaw that you should be aware of before installing (see below).

Here's what got better:

• Simplified the way in which tag properties are prepared (simpler regexp, reduced code repetition).
• Removed a misleading comment.
• Added some (hacked in) support for scoped tagging -- that is, generating tag clouds based on descendants of a page rather than an entire wiki. Courtesy of David Antila, who is looking to expand this in the future. You can get at it via the new method Tag.forDescendants(), which takes a path as an argument.
• Fixed a bug which prevented you from deleting all of a page's tags.
• Fixed a bug which would totally break the pivots (i.e. finding other pages with given tag) if you had any pages where the tags were stored as a 'string' type property (rather than a 'stringList'). (But see below!)

Here's what got worse:

• The last bullet point above.

This new version will produce the wrong results in the following circumstance: you have some tags which are substrings of other tags.
For example, in this new release, pivoting on "test" will also return pages tagged "tests", or "testing". Pivoting on "dog" will return pages tagged "dogfood".

Basically, we are headed towards a tagging library which seamlessly accepts both string and stringList properties, both single and multiple word tags, and Just Works. We aren't quite there yet -- this is a development release along the way. If the caveat above looks like it will be a problem in your wiki, you may want to wait until the rest of the fix is in. If you need other stuff from this release in the meantime, but can't accept this flaw, install and then roll back your pivot pages a revision, or email me (phil@unimedia.org) for help figuring out what needs to be tweaked. This will only work if your tag properties are all stringLists.

Back to ProjectBlog