[TYPO3-dev] turkish caseshifting in core

[Gür Yildirim]

Hi Elmar,

I agree to all you've said ...

I tested <LINK> with a dotted I and it still doesn't work. Looks like lower-dotted i is the only way it works, no matter tr_TR, TR or tr_TR.utf-8

I tried different combinations, please take a look at http://t3dev.ekcon-it.de/index.php?id=1
There you can test the problems on your own. It's ugly but with *TR* LINK and link work so RTE for instance won't have problems with that.

But: since the only locale available for turkish is *tr_TR* output of the modules for dates and other stuff is english. So I did this:
  cd /usr/share/i18n/locales/
  cp tr_TR TR
  rcapache2 restart
  Clear the cache and reload
But unfortunatly this didn't work too.
With *tr_TR* date-outputs are turkish but with wrong encoding.
With *tr_TR.utf-8* date-outputs are turkish and fine.

Getting the focus back to the LINK-link-inconsistency there is one thing to mention:
There is one non-logical thing with caseshifting in turkish:
English words in turkish are written with non-dotted-upper or dotted-lower i ...

To make confusion complete about not working dottet-"LINK" and working "link" I want to finish my posting citing my posting in the bugtracker:

>>
I do add my note here, although it does extend the view-ankle a bit. I did post my problems belonging to turkish language to news-lists, both german and english, but none could give me some advise, so:
Indeed, the "I" is causing troubles, since obviously typo3 seems to lowercase all markups in the forms internally for instance. I'm not into the code just atm. so please accept these "tutorial-meant" examples:
When there is a mark for a form-label like ###YOUR_CITY### in my template which might look like this in the output (en, de, tr): "Your city:", "Ihre Stadt", " ".
So I go back to the template and change the mark like this: ###YOUR_CiTY###. Result: "Your city:", "Ihre Stadt:", "Kentiniz:".
But in cases of marks containing more than one "I" this really starts getting "stupid", because it ends up in a *quiz*: I have to test if for example
###TRY_IT_AGAIN###, ###TRY_iT_AGAIN###, ###TRY_IT_AGAiN### or ###TRY_iT_AGAiN### makes my label appear.
Why should lower or upper make a difference? Turkish characterset is similiar to the american character set but has some additional charachters. In this case I just mention the '&#305;', which is written as 'I' n uppercase, *not* being the same as 'i' or '&#304;'. You see the disaster we face?
<<
Note:
&#305; = dotted-upper-I; &#304 = non-dotted-lower-i

-----Original Message-----
From: typo3-dev-bounces at lists.netfielders.de [mailto:typo3-dev-bounces at lists.netfielders.de] On Behalf Of Elmar Hinz
Sent: Saturday, March 25, 2006 7:25 PM
To: typo3-dev at lists.netfielders.de
Subject: Re: [TYPO3-dev] turkish caseshifting in core

Hello typoniks,

dotted and undotted Is in turkisch.

I don't know if I can help to fix this problem, but I tried to approach this interesting problem by studying

/usr/share/i18n/locales/tr_TR

on my machine. Because probably few people will do this, I report a summery of my results.

To say it short:  turkish has the disadvantage to be very sytematical with Is.

There are 4 Is: dotted and undotted both in higher and lower case.

International languages only know 2 of them and switch between

dotted-lower <=> undotted-upper

turkisch knows all and switches consequently:

dotted-lower <=> dotted-upper
undotted-lower <=> undotted-upper


Gür reports that only lower-dotted i works for <link>.

Undotted upper does't work. It will by lowercased to the lower-undotted wich isn't used in <link>.

Taking it logically the dotted-upper should be converted to dottet-lower in turkish. Please Gür test to write <LINK> with dotted-upper I.


Martin Kutscher has proposed two solution to switch between POSIX for evalution of tags and locale for evaluation of content bewteen tags.

PROBLEM: case conversion and Turkish


A quickhack would be a regular expression during parsing the code to make links from <link>.  something similar to '/l.nk/'.

Regards

Elmar





Gür Yildirim schrieb:
> Hi Core-Developers,
> 
> I guess a serious Gremlin will make its way straight to T3_4.0 Let me 
> try to scratch what happend:
> 
> 22.07.05
> Metin Yilmaz reported at http://bugs.typo3.org/view.php?id=1303 (ID 1303) a bug about troubles with config.locale_all=tr_TR and broken links.
> I was new to typo3 and didn't ever read the mailing lists.
> 
> July/August of 2005
> I advised our editors not to use RTE but to write links by entering 
> <link ...>...</link>, Which I found out to make links work. 
> UPPERCASE-notation fails (forget about ts_ref-notation)
> 
> 08.12.05
> I reported my own experiences (anger) as a Note at http://bugs.typo3.org/view.php?id=1303. Several people since then asked me by direct mail if there was any development according to this case, but I was sorry to decline.
> 
> 09.12.05
> Martin Kutschker, whom I am very thankful for, posted his thoughts 
> about the origin of the problems, which now seem to be obvious to me 
> ;) to typo3-dev (Subject: "PROBLEM: case conversion and Turkish"). A 
> list I wasn't reading those days so I missed to answer or comment just 
> in time. Nevertheless I wondered why there was no changes to the 
> bug-status (no new notes, monitors, etc.)
> 
> 25.03.06
> Wanting to meet trouble halfway to t3_4.0-release-day I checked 4.0rc1 
> and ... here we go again ... :'(
> 
> Can anyone help me out by saying everything's gonna be alright?
> 
> Gür "Cenk" Yildirim
_______________________________________________
TYPO3-dev mailing list
TYPO3-dev at lists.netfielders.de
http://lists.netfielders.de/cgi-bin/mailman/listinfo/typo3-dev