search function weirdness

[tanuki]

Hello!

I noticed this some time ago, but I never asked.

Right now I was doing a search with the site's search function in order to see if there's already been a thread on 〜ずつ. I went to "search" on the left menu, typed "ずつ" into the box and chose "Forum posts".

What I get is a text saying "114 Forum Posts found matching search criteria" and under that there are two links to the shiritori thread. I go to page two, in order to find the rest of the found posts and there are no links there! I go to page three and there are six links. I go to page four and there is just one single link...well, you get the idea.

Does this happen to everyone? The only thing that I can imagine is that the posts are ordered by date and that each page represents a certain time span, but that doesn't make much sense...

Thanks in advance for your replies, bye!

[zengargoyle]

the PHP-Fusion search is broken beyond repair when searching for non-ASCII text. Fusion was never designed to handle non-ASCII. it's searching on the bytes that make up the text: 0x82 0xb8 0x82 0xc2

when it finds matches and does its counting it uses functions that don't handle non-ASCII text and things get all confused. it's a hopeless situation.

so, if you search non-ASCII and find something... you got lucky.

[tanuki]

It looks like I'm not getting lucky tonight...

*ahem* I mean, thanks for the explanation! Does that mean that the only way to be sure that the search function is going to work is writing in romaNji? Aaaaaaaah! We are surrounded!

[zengargoyle]

pretty much.

there's the possibility of rewriting the search function to properly handle the SJIS nature of this site, pretty much has to be a complete replacement of the search function.

sorta on the backburner is somehow converting the site to use some other software. high on my list is Drupal, but there's a bit of hangup on converting the old entries from SJIS to UTF8 (sorta hard), but other software (like Drupal) has been written from the start to be fully UTF8 compliant. this is sorta why i'm loathe to spend lots of effort hacking around the Fusion software. in the end it seems better to move off of Fusion to something that supports UTF8. maybe one of these days....