> Okay, let me try again: This has nothing to do with regex matching in code; > I am using the built-in full-text search index to find and group my tags, > hence they need to be words that are recognised by the indexing function. > So, by hacking the indexer I allow '&' as a indexed word character and that > way words like &tag show up in the index. Then it is very easy to create > tag clouds, tables of tags with matching pages and search for pages > containing multiple tags. After all they are just 'word' with a '&' > prefix... > > It works very well as I said, so I was looking for a way to achieve the same > feature without hacking, hence the idea of tripling the first letter: it was > just another method of creating unique tags (distinct from other normal > 'words' in the content). The plugin I created hides the '&' symbol when the > tag is displayed, I would do the same with the triple letter. > > Does this make any sense yet? I could link to some screen shots if that > would help! Got it now. Make much more sense now. Maybe the easiest way would be to have a way to modify what the indexer takes as word boundaries from within a plugin? I like this approach to tagging by the way. Andi -- splitbrain.org -- DokuWiki mailing list - more info at http://www.dokuwiki.org/mailinglist