Textpattern CMS support forum
You are not logged in. Register | Login | Help
- Topics: Active | Unanswered
Re: Better (?) <txp:link_to_prev/next />
jstubbs wrote:
That’s funny, changing the status of an issue to invalid renders it invisible, so I never saw Jukka’s reply.
The issue should stay intact no matter what the status is changed to. By default the issues list shows only open issues. The search form at the top can be used to filter it. May not be terribly intuitive, but that’s how Google code’s issue list works.
Last edited by Gocom (2013-01-22 08:19:09)
Offline
Re: Better (?) <txp:link_to_prev/next />
Actually, quicksort is log(n) in memory, but n log(n) in time, so scaling should not be a problem if there is enough memory.
Gocom wrote:
That’s not terribly good.
For sort="LastMod desc" I get: core=0.007s, plugin=0.0044s, with additional benefice of link_to_first/last. :P
Edit: maybe, it’s possible to combine both approaches, checking existing indexes (SHOW INDEX FROM textpattern).
Last edited by etc (2013-01-22 09:34:38)
Offline
Re: Better (?) <txp:link_to_prev/next />
etc wrote:
It would be nice to have a test txp install at hand, are there any?
Try wet_lorem_ipsum to create any number of dummy articles.
Offline
Re: Better (?) <txp:link_to_prev/next />
wet wrote:
Try wet_lorem_ipsum to create any number of dummy articles.
Nice toy for this week-end, thanks! I will try to fork to the default link_to_next if EXPLAIN uses indexes. Are there any changes in link_to_next in 4.6 to keep in mind?
Offline
Re: Better (?) <txp:link_to_prev/next />
Done some tests on 10000 articles sorted by LastMod using filesort. Retrieving the whole list is about 15-20% faster (0.065) than generating two prev/next links (2*0.04). Something (array_search or dirty code?) in the plugin adds extra 0.01s, but it is still a little faster than core. They don’t benefit from each others cache, of course. I have tested array_search on range(1,10^6) without any problem.
Last edited by etc (2013-02-05 09:03:33)
Offline