Yes it’s doable, but it is hard to make it work across all different versions and all different configurations. A lot of other people are using Textpattern happily with all kinds of languages. And just because another package works out of the box with the setup you have, doesn’t mean they get necessarily get it right in any possible setup and what works for you might break for someone else.

We are working on it.

marios · 2005-07-26 00:59:55

What would be highly appreciated,would be some test results with some different server environement (with greek hosting companies,if you prefer,but not necessarily).
Thanks again , Sencer for your invaluable help.

with best regards, marios

spiros · 2005-07-26 10:24:17

> Sencer wrote:

> Yes it’s doable, but it is hard to make it work across all different versions and all different configurations. A lot of other people are using Textpattern happily with all kinds of languages. And just because another package works out of the box with the setup you have, doesn’t mean they get necessarily get it right in any possible setup and what works for you might break for someone else.

>We are working on it.

Hi Sencer, it is great to know that you are working on it and you take globalization seriously!

I have started a research on unicode and multilingual content support across a variety of CMSs and I am planning to write an article about it. However, I will delay it because I think TXP is just a great tool and should be included in the list!

to marios > Cannot be interpreted as UTF-8 >Allso,if something validates, it is not necessarily correct XHTML. >(And then you don’t want those diamond shapes on your page allso , I guess)

Marios, I am not sure what you mean with “Cannot be interpreted as UTF-8”. I have UTF-8 encoding in my browser and it works and I see no rhomboid shapes either (with mozilla that is). I just tested it with IE though and I get square shapes instead of the accented polytonic characters and some of the Chinese ideograms do not display properly…. oh well c’est la vie!
However, standard Modern Greek looks fine with UTF-8 in both browsers.

Sencer · 2005-07-26 15:18:29

Can the people please send me (email) the result of this query:

SHOW variables like 'character_s%'

(you should do this in the database, where txp is installed, please) and tell me in a few words, the problems they are seeing on a default install with no changes made.

Thank you.

spiros · 2005-07-26 18:35:49

Here you go Sencer

character_set latin1
character_sets latin1 big5 czech euc_kr gb2312 gbk latin1_de sjis tis620 ujis dec8 dos german1 hp8 koi8_ru latin2 swe7 usa7 cp1251 danish hebrew win1251 estonia hungarian koi8_ukr win1251ukr greek win1250 croat cp1257 latin5

Problems in default install:

Greek characters converted to html entities.

Last edited by spiros (2005-07-26 18:36:42)

marios · 2005-07-27 18:34:57

Sorry ,for not being here, I went on a small trip, and come back (well,with some little) fresh energy, unfortunately I’m not that confident with mySQL, so again this is on the lang table, not sure though.On rv 635

character_set_client utf8
character_set_connection utf8
character_set_database utf8
character_set_results utf8
character_set_server latin1
character_set_system utf8

(didn’t see any problems)

with best regards, marios

Last edited by marios (2005-07-27 18:48:43)

marios · 2005-07-27 18:56:52

Wonderfull, Sencer

thanks again for everything ,

with best regards, marios

Last edited by marios (2005-07-27 18:59:47)

Sencer · 2005-07-27 19:22:38

I have been working on solving the problems with mysql4.1 and i18n issues. I have commited my changes in rev636. marios, can you please do a fresh install and see wether you have any more issues? I hope you don’t. On my tests everything worked, inluding updating language via RPC. Text on Frontend and Backend displayed correctly.

Explanation: What the changes do is they create utf8 tables if we are on 4.1 and up. They add a config-value to config.php ( $txpcfg['dbcharset'] = 'utf8'; ) and the connection charset is set explicitly to this value for each connection.

For older installations: If (and only if) you are sure that your tables are already created as utf8 (because you manually did that, or your DB defaulted to that) then you can add the previous line to config.php and thus “upgrade” an existing installation (of yourse you also have to replace all other files as well, just do a regular upgrade).

I have also added some stuff to diagnostics when you set high level – there should be a couple of mysql/charset related lines.

@spiros: You are on mysql4.0 (or older). This is not a grave problem, since you can still write and display your posts, it’s just that the db won’t correctly search and order in posts. There are no problems with i18n on mysql4.0 and older, when using latin1-tables.

> Greek characters converted to html entities.

If this happens on the front-end, than this is actually by design. Because not everybody uses utf8 for their websites, Textile converts all characters that are > 0×7f to html entities but only for rendered posts. This is not strictly “necessary” if you are using utf8 headers on the webpage, but it doesn’t hurt either.
Though I understand the desire that Textile behaved differently and we’ll see what we can do about that.

btw: WP will only work correctly if you have 4.0 and lower or the “right” defaults for 4.1 and up. They “get it right” by simply letting it default to whatever and not caring about the db, if I didn’t overlook anything.

marios · 2005-07-27 20:44:54

I have been working on solving the problems with mysql4.1 and i18n issues. I have commited my changes in rev636. marios, can you please do a fresh install and see wether you have any more issues? I hope you don’t. On my tests everything worked, inluding updating language via RPC. Text on Frontend and Backend displayed correctly.

Bingo, verified using 4.1.11 and r637

Sencer, you are such an incredible geek,so how can I pay you back then, at least I owe you a case of beer.
I tell you what, Ill take a break now,and get myself some at the supermarket and toast to your health and TXP,s well-formedness,

With best regards, marios

Last edited by marios (2005-07-27 20:48:08)

Sencer · 2005-07-27 21:04:55

I’m glad it’s working for you! I have to attribute most of that to mamash, zem and you though, since I only put together and bring to an end what all three of you have been figuring out. ;)

marios · 2005-07-27 22:06:43

Well, the geek badges go to you then, Zem and mamash ,

So what about me then, errr,
no good, keep on learning, learning, learning under some sort of circus circumstances.

with best regards, marios

Last edited by marios (2005-07-27 22:07:54)

spiros · 2005-07-27 22:48:05

Marios and Sencer,

Could you please share with us the results? I.e. UTF-8 / Windows-1253 works without creating html or numeric entities and displayes without squares for both IE and Mozilla? Could you also share the latest rev so that I could test its behaviour on my server?

Thanks!

Last edited by spiros (2005-07-27 22:49:35)

Sencer · 2005-07-27 23:14:17

spiros, you are on mysql4.0 or lower, you won’t be seeing much of a difference I suppose. As I tried to explain the creation of the entities is not a bug or error or a necessitiy – it is currently done by design of Textile (the text-formatting used in textpattern).

Though you should not be seeing squares or anything similar (I haven’t after I tried it out on mysql3, mysql4.0 and mysql4.1 with latin1 and utf-8 DB. Always used utf-8 headers on the webpages though.)

If you want to try it out, see here: http://www.sencer.de/code/rev638.zip (temporary, will likely die soon).

Textpattern CMS

Textpattern CMS support forum

#16 2005-07-26 00:30:41

Re:

#17 2005-07-26 00:37:31

Re:

#18 2005-07-26 00:45:04

Re:

#19 2005-07-26 00:59:55

Re:

#20 2005-07-26 10:24:17

Re:

#21 2005-07-26 15:18:29

Re:

#22 2005-07-26 18:35:49

Re:

#23 2005-07-27 18:34:57

Re:

#24 2005-07-27 18:56:52

Re:

#25 2005-07-27 19:22:38

Re:

#26 2005-07-27 20:44:54

Re:

#27 2005-07-27 21:04:55

Re:

#28 2005-07-27 22:06:43

Re:

#29 2005-07-27 22:48:05

Re:

#30 2005-07-27 23:14:17

Re:

Board footer