Mario, Yes, I specifically confirmed with Jeremy that the target set of tiddlers could be limited, because of this performance hit... My wiki is not small (>800), but the demand for freelinking is really only about my vocabulary terms and proper name entries.) I am happy with a workflow that piggybacks on tags since they line up neatly with why I want freelinks... but a specific field-based variable would be fine too, and would help us interrupt nuisance freelinks to "will" (say) without having to untag it as a definition.
It's clear that the challenge with aliases is that it's a burden on processing. I'm still a bit astonished about this, since TWC had all of this desired behavior available (RedirectMacro), including with alias fields transparently woven in; it's only because TW5 is so much more efficient at *other* things that it can't efficiently handle this challenge. One low-hanging fruit would be to allow freelinks to be case-insensitive (or even just insensitive to Proper vs lower) so that one frequent cause for false negatives in ordinary pasted text is easily addressed. But then singular/plural variants are a close second, and then there are many other potential reasons to seek aliases of the kind that only the author of a wiki can anticipate... Cheers, and thanks for helping to think through it! -Springer On Wednesday, April 22, 2020 at 2:57:30 PM UTC-4, PMario wrote: > > On Wednesday, April 22, 2020 at 7:15:30 PM UTC+2, springer wrote: >> >> Getting an alias field to play well with freelinks is THE DREAM! >> > > hmmm .. ok > > >> To be clear: something like uni-link helps one kind of use-scenario -- >> the one where a single author is generating tiddler content, and is aware, >> while typing, of where there's an alias available. >> > > That's a point > > >> My own case (perhaps like OP here, blarg H) is that I'm often pasting in >> text from elsewhere, and I just want all the resonances to other tiddler >> titles to self-recognize. >> > > A good point too. > > >> Jeremy's freelinks macro is GREAT for that! But simple things like >> capitalization and plurals get broken. And since I'm pasting in text from >> elsewhere (sometimes in a formal-citation context) it's really not >> appropriate for me to mangle the text itself. >> > > OK > > >> I'm in no position to rewrite any plugins (least of all now, given my >> day-job). But I would actually pay real money for this work. And I'm not >> rich. :) >> >> -E Springer >> > > Hi, > Sometimes it needs posts like this, to broaden someone's (my) view, where > it is too narrow-minded. ... > > But ... I still have a problem. It's the potential performance hit with > many 1000++ tiddlers > > As Jeremy pointed out in the freelinks plugin description, the mechanism > used is heavyweight, compared to the "normal" text parsing. Alias-linking > is also relatively heavyweight, compared to standard links. It's similar to > "backlink" handling. > > Both mechanisms use internal caching, to speed up lookups. But we still > have to rebuild this cache very often. This means we need to touch every > single standard-tiddler, to build the cache. > > So from my point of view, we need to make more tests with a huge amount of > tiddlers, tags and aliases. With and without the plugins enabled, to be > able to measure the "real impact". > > My second concern is, about the "false positive" links that can happen if > a tiddler is named "test" or "task" or "in" or "out" ... you name it. So I > personally would prefer a > > \define tv-freelinks() yes > > at the head of a tiddler, instead of globally enabling freelink handling. > ... This doesn't reduce the time to create the cache, but it would limit > the "false positive" links. > > Thinking ... ;) > > have fun! > mario > -- You received this message because you are subscribed to the Google Groups "TiddlyWiki" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To view this discussion on the web visit https://groups.google.com/d/msgid/tiddlywiki/37198464-df78-490c-ba77-a70155253950%40googlegroups.com.
