Op donderdag 16 februari 2006 17:11, schreef Jeff Eaton:
I use it on my blog, but to be honest I think the current path system makes pathauto unusable for core inclusion.
A more robust algorithmic url-aliasing mechnism, like the one that chx and company brainstormed a few months ago, has a lot more potential IMO. On large sites, auto-generating aliases creates unmanagably huge path tables, colliding urls with ugly _0, _1, _2 suffixes all over the place, and so on. That's not a failing of pathauto, just a testament to how much it DOES do in an alias scheme that's not well suited for some of the things it's often used for.
I second this. Its usefull for sites under the 2-3000 nodes (blogs etc) but for anything bigger it is not usefull. And combined with freetagging, the data it needs it grows into a monster. I tested it on a big site (10.000 nodes/900 users/2000 tags) and the alias table contained around 100.000 lines! For drupal .org that would translate into something like > million aliases :). Loaded into cache on each page load? A -1 for inclusion in core as it is. We need the *functionality* in core, but certainly not in this particular imlpementation. We talked about tokenised menu system. Making that system work two ways (path->callbacks AND url()->paths) is the route to take IMO. Storing 20.000 blogs/My_user_name/feed is not a good idea. Storing one blogs/%username/feed that is parsed/cached for earh url() or so is a better idea. And no, I am not volunteering for this, but that fact should not rule me out of the -1ers for this one in core. :) Bèr -- | Bèr Kessels | webschuur.com | website development | | Jabber & Google Talk: ber@jabber.webschuur.com | http://bler.webschuur.com | http://www.webschuur.com |