Khalid B wrote:
What Ken Rickard said makes me think of another solution:
If we provide a generic way to exclude certain paths, that can cut down the processed aliases.
For example, a site may only want to alias "taxonomy/term/*" and "node/*" and nothing else.
Another site may only want aliases for "node/*" and "user/*"
The code this way is generic and customizable on a per site basis with no patches, ...etc.
You get the idea ...
Interesting idea. We could look at the system urls that actually are aliased on a particular site, try to abstract some regexp from them (mainly by replacing numbers by a regexp matching numbers, I'd think. Maybe also by truncating them.) and store the resulting regexps in a variable. Then, before we'd look for a matching alias for a particular system url we can first match that urls against our regexps. I somehow doubt that this would be faster, though. Cheers, Gerhard