Thanks for your answer, andi!
andi wrote
How? Is it real text or random gibberish?
All the materials are real text from my database(a dictionary like database). I used a PHP script:
Firstly, My script grabs /doku.php?do=search&id=sometopic to get a right page name of sometopic.
Then it writes a txt file with the right page name, directly to /data/pages in wiki format.
Then it visits the page once /doku.php?id=somwtopic
Finally, it visits /lib/exe/indexer.php with debug mod open.
Is it OK?
andi wrote
/bin/indexer.php would be the better way.
Thank you~I didn't know how to index wiki manually, and found a topic here:
http://forum.dokuwiki.org/thread/775, so ...:)
andi wrote
Exactly how slow is it? What kind of queries did you test?
I searched a word which is included in about 3,000 pages, this word is only in page(need fulltext search), then the wiki just appeared the search animation(loading.gif). About an hour later, I closed the browser.
Other searches that the keyword is included in the page title are much faster, they need less than 1 min to show the matched page titles, but the fulltext search section still shows loading.gif for ever.
Regards,
George