... in the page title, and in the body: äöüßÄÖÜ
Other accents too:áéíóú
RS adds all of the iso8859-1 characters above 160, for even heavier testing:
160: ¡¢£¤¥¦§¨©ª«¬��®¯ 176: °±²³´µ¶·¸¹º»¼½¾¿ 192: ÀÁ����ÄÅÆÇÈÉÊËÌÍÎÏ 208: ÐÑÒÓÔÕÖ×ØÙÚÛÜÝÞß 224: àáâãäåæçèéêëìíîï 240: ðñòóôõö÷øùúûüýþÿ
KBK edits the page to see whether the characters get damaged in the round trip.
mjk I can promise, that in this Wiki will visit some browsers, that will convert ISO-8859-1 characters to UTF-8. You can find my little rant from my user information page. Unfortunately, I'm yet unable to identify, which browsers change character encoding (I have tested with Mozilla 1.6, Firefox 0.8 and IE 6.0).
Let's test links: testing umlauts: äöüßÄÖÜ... Moo!
rmax I've investigated that UTF-8 conversion problem today (that's why I created this test page). The current state is, that it seems the modification came from within wikit or tclkit, because some pages were converted from iso8859-1 to UTF-8 twice without being edited by anyone.
jcw - I probably caused this by some sloppy pack/rename action over the past several months, though I don't recall having done anything that disruptive. For now, given the way this page seems to hold out, I'm assuming things will not deteriorate further.
This is running Tclkit 8.4.5 (static) on Linux, btw. When launched from the command line, that build tells me the system encoding is iso8859-1.
Best we can do now I guess, is to manually fix whatever turns out to have been mis-encoded (there's CVS history to help figure it out, if needed). If there is ever any further regression, please let me know - that would mean there are still problems. Reading mjk's comments above, I cannot rule out that some browser is causing the incorrect re-encoding, though it cannot explain how things can go wrong with no intermediate edit, or how a page title got mangled. Maybe some page changes never made it to CVS. Hmm... it's not really clear.
When saving this page, it says "page saved <gibberish>", then the modified page comes up and it looks ok again. Hm, encodings are still not 100%. -jcw