Hi, if I try to commit a website with ISO8859-15 umlauts (like öä) I only get an empty page as the result. If I rebuild from hand I get funny things like: Malformed UTF-8 character (unexpected non-continuation byte 0x68, immediately after start byte 0xfc) in substitution (s///) at /usr/share/perl5/IkiWiki/Plugin/camelcase.pm line 18. If I switch the page to UTF-8 with recode ISO8859-15..UTF-8 I get something like a 2byte char splittet into to single byte chars. But beside of this problem: Thanks for ikiwiki!! I love it Alex
Alexander Wirt wrote: ikiwiki requires all input pages be encoded in utf-8. Can you send me an example page in utf-8 that canses a problem?
Joey Hess schrieb am Mittwoch, den 07. Februar 2007: Hi Joey, *snip* Which is no problem, but it should provide more informations than an empty page. Attached is the UTF-8 file I tried. Alex
Alexander Wirt wrote: Well, it did provide helpful ugly perl warnings about failure to parse the encoding as unicode. :-) Help me see the problem with ikiwiki's handling of that file? If I view it in a unicode xterm in vim, I see it contains the word "GefÃŒhle". I'm unsure if this is the correct rendering -- is it supposed to contain the letter à followed by the letter Œ? These are not German letters I'm familiar with[1]. It seems that it still might not be encoded properly in unicode in the source file. ikiwiki renders this to the same thing, "GefÃŒhle", which is just a case of garbage in, garbage out AFAICS. If I change the file to contain umlauts like öä, these are rendered fine in the result.
Joey Hess schrieb am Donnerstag, den 08. Februar 2007: Yeah, I understood the error message, but it would be more helpfull to have them the browsers output instead of an empty page ;). Maybe its the fault of recode, I will do some more tests. Maybe I am able to find a unicode system. Normaly it should get rendered as Gefühle. You can ignore the utf-8 part of the bug for now until I have some more news. Thanks Alex
Alexander Wirt wrote: Yeah. I don't know offhand of a way to do that though. Perl handles parsing the input as utf8 itself, and does the warnings too. It might be possible to wrap some code that forced the warnings in an eval and show the error to the user via that, but I'm not really sure how to do that.