[pmwiki-users] FW: PMwiki and EBCDIC coding

[Petko Yotov]

On Thursday 07 May 2009 14:49:41 marc at spookymonster.com wrote:
> >> led to the creation of PUE()? If I knew that, I might be able to match a
> >> similar EBCDIC-friendly regex. Any info would be greatly appreciated.
[...]
> I tried that, but the PUE description was a little too vague. In
> particular:
>
> "url_encode also encodes '%', '/', and other ASCII characters, which often
> isn't what we want."
>
> What characters specifically don't we want to convert, and why? Does PMWiki
> use them as delimiters in some other portion of the code?

Hi. "/" may be used as delimiter of Group/Page, and in the case of CleanUrls, 
it shouldn't be url-encoded. Also, sometimes a string passes twice through 
PUE(), and should not be double-encoded-- that is very important in 
international wikis with characters in PageNames different from [A-Z0-9] 
which get encoded like "%d0%9f".

I suspect you may be able to fix the encoding with a special script or a 
setting like PmWiki's own scripts/xlpage-*.php, but I have absolutely no way 
to test it in EBCDIC. And maybe restrict group/page names to A-Z0-9 
characters ($NamePattern, $GroupPattern).

Thanks,
Petko