German characters in the wiki

[Gryzor]

Hello guys,
Toyday Morn told me that he can't access http://www.cpcwiki.eu/index.php/Das_große_Grafikbuch_zum_CPC . Well, Opera gives me a Not Found Apache error, Chrome gives me another error, but *very* interestingly, my Android device opens the page just fine! It's obviously down to the special character, but any advice?
Cheers
t

[MaV]

Difficult to debug:

The letter ß is Unicode U+00DF and %C3 %9F. If I type the latter into the http line instead of the "ß" the error remains the same.

The error message itself suggests bad coding:
"MediaWiki internal error.
Exception caught inside exception handler"

An exception that's caught but the code that catches it doesn't do anything meaningful to explain the situation.
Interestingly your link in the message stops exactly at the letter ß and interprets the rest as normal characters.

My guess is there is some switch in the wiki that changes the interpretation of the http requests of various browsers, or the interpretations of the encoding (like utf-8/utf-16, ISO-8859-1), or perhaps the wiki now just works with a different encoding, while some browsers silently assume another.

Are there more links that use European characters which have problems? Links containing among others áàèéöäüÁÀÈÉÖÄÜçÇ? That should help you with testing, and give you a hint what is wrong.

Since it works on the Android device, I don't think the data itself has suffered from the server move.

[Morn]

Well, this problem seems to happen for all pages that have any German umlauts in the page title. Just look at the German section of the book list. Everything that contains a "für", "große", etc. produces the same error message in Safari and Firefox for me.


Martin

[Gryzor]

Yeah... we knew about the umlauts and thought it had to do with some screw-up during the first server change. But now, since I saw that it actually works with some set-ups (like the Android browser) it seems there's something else at play... I'll try at the mediawiki community.

[Morn]

I've tried it with Opera and Links, too---same error message. I'm not sure why the Android browser acts differently.


It's a bit strange that the browser type would make any difference. They all encode non-ASCII characters in URLs in the same way, I think.


Martin

[Bryce]

I get the same error. Can't we rename the pages with "grosse" instead of "große" and "fuer" instead of "für" etc..
There can't be that many pages.

I does seem to be a wiki error though. The link above stops (isn't underlined) from the "ß" onwards. But links with Umlaut/ß on other sites are shown properly and Firefox opens them perfectly.

Bryce.

[Gryzor]

Can someone who has set their computer locale as German give some feedback? Do you get the same errors?
@Bryce: we had done it with a couple of articles - basically you can't rename an article but you can create a new one... but since it seems it's not related to the db move I should try and fix it...

[Bryce]

My PC is German (Keyboard) with a German Win7 installation and German Firefox.

Bryce.

[Gryzor]

Not sure if all this has a bearing; is locale set to German? Maybe it don't matter, but you never know with Windows, since the early days of manually editing config.sys to make it properly display Greek has left its mark...

[Morn]

Not sure if all this has a bearing; is locale set to German? Maybe it don't matter, but you never know with Windows, since the early days of manually editing config.sys to make it properly display Greek has left its mark...

All affected pages worked perfectly on the old wiki server. I don't think a client-side config.sys can help when it looks like the server is misconfigured in some way. Are you sure those pages in the Android browser weren't cached?


Martin

[MaV]

My computer is set to German as well, as it should be.

So every input to this thread were on German computers.

And Morn is right. They did work properly on the old server, so it's a server side problem.


And as bryce and I have found out: Even the link in your original posting stops at the "ß" and interprets the characters from then on as normal text.
Again, it's best check the wiki options, or the options of the underlying scripting language (PHP?).


Do any characters of other European languages suffer the same problem here (French, Greek, etc.)?

[Morn]

Do any characters of other European languages suffer the same problem here (French, Greek, etc.)?

French books with accents in the title also produce the error. So do Spanish titles with that wiggly line over the n:
(That last word is "Ninos", with an accent over the n, in case the forum software cuts it off.)


http://www.cpcwiki.eu/index.php/El_Amstrad_y_los_Niños


Martin


P.S. As before, the link automatically created by the forum software stops at "Ni"!

[Bryce]

P.S. As before, the link automatically created by the forum software stops at "Ni"!

Is that like "The knights that like to go Neee" ?

Bryce.

[Morn]

Is that like "The knights that like to go Neee" ?

Bryce.

Yes, we seem to have a Monty Python virus on the new server. 


Martin

[nurgle]

Arghhh, Umlauts always get back to bite me. Can be the locale system setting, MySQL locale setting or apache setting. I will try to solve this. Thanks for reporting.

[Bryce]

Yes, we seem to have a Monty Python virus on the new server. 

Maybe we just need to bring the server a Shrubbery.

Bryce.

[Morn]

Maybe we just need to bring the server a Shrubbery.

Bryce.

Or, better yet, combine the shrubbery with the Killer Rabbit of Caerbannog and the Holy Hand Grenade of Antioch. And then nuke the site from orbit--it's the only way to be sure. 


Martin

[Morn]

Arghhh, Umlauts always get back to bite me. Can be the locale system setting, MySQL locale setting or apache setting. I will try to solve this. Thanks for reporting.

I would suspect PHP. Both MediaWiki and the forum software use it, and the fact that automatic link creation in the forum fails as soon as it encounters a non-ASCII character strongly suggests a PHP configuration issue I think.


Martin

[Gryzor]

PS Morn: good thinking...

[robcfg]

I've also noticed that there are two pages for that book on the wiki. I don't know if that could cause any issues.


Search for Grafikbuch and you'll see it.

[AMSDOS]

PS Morn: good thinking...

LOL! I just brought & watched the Holy Grail for the first time the other day! Spooky! 

[Gryzor]

Must've watched it a dozen times, fantastic movie...

[nurgle]

Just a quick update: The database is mixed latin1 and utf8, so it's not that easy to debug. I tried to convert everything to utf8, but that fails since some indexes can't be created properly then. Setting the charset in php.ini or on a database level did not help. So, there is not solution yet. I am sure I will find one, but it may require some more thinking plus a lot of trial-and-error. I am working on it.

[MacDeath]

ß

Well, why can't you just SS like the rest of the world....

Oh yeah, last time Germany used the SS, it didn't ended that well.

Ok, I guess the post-war germany imposed the use of the ß because all Nazi related stuff was forbiden so Waffen-ß was more politicaly correct then waffen-SS...



Sorry, was just kidding...

Btw I love Waffleß and LuftWaffles (with Nutella of course)

[Morn]

Well, why can't you just SS like the rest of the world....

Oh yeah, last time Germany used the SS, it didn't ended that well.

Ok, I guess the post-war germany imposed the use of the ß because all Nazi related stuff was forbiden so Waffen-ß was more politicaly correct then waffen-SS...



Sorry, was just kidding...

Btw I love Waffleß and LuftWaffles (with Nutella of course)

We simply need our umlauts because they are officially cool: http://en.wikipedia.org/wiki/Metal_umlaut


It's a little-known fact that umlauts are so cool that German houses don't need air-conditioning even in August. I actually have to wear a parka and snowshoes right now.


Martin


<<<Typical German summer weather