On Donnerstag, 13. Dezember 2007, cnit wrote:
> > Well, not sure ... in general, SMW and MW have rather similar ways of
> > creating tables, and normally they just come out having the same settings
> > (it does not matter if they use Swedish collation, as long as they use
> > the same).
>
> Collation does matter for string ordering operations (which are used
> by MW, for sure). Using Swedish collation with Cyrillic texts surely
> would lead to catastrophe, just ask any of Russian mysql developer.
> Even with German language I really doubt that Umlauts and so on
> have the same codes in different collations.

Oh, of course. But it would not give you DB exceptions, just unexpected 
ordering results (and, yes, for Cyrillic this truly is an issue).

But finishing this issue: setting proper default collations for your database 
before installing MediaWiki or SMW should usually work as well. But I am not 
een sure whether MW has other direct means of choosing the collation, or 
whether it just leaves this to the DB.

>
> > I believe that the observed problems stem from having a database where
> > settings somehow changed between MW and SMW installation, so that SMW
> > used other default collation settings than MW used when making its
> > tables. I assume it should be possible to have a script that just changes
> > the collation of all text-fields in all tables to whatever value some
> > admin wants.
>
> It may be possible even though probably requires re-indexing.
>
> > Note that the UTF8-DB-encoding option (referring to the one offered
> > during install now) is considered experimental for MediaWiki.
>
> Hmm.. AFAIK only UTF8-binary is considered experimental, not
> utf8_unicode_ci. And MW surely uses UTF8 throughly it's core - it
> doesn't even work without PHP mb_string extension.. How would the
> mysql server suppose to store Cyrillic or Chinesse characters
> in a table with a database charset that doesn't even have such
> character codes defined! And with UTF8 posts you may enter these
> characters any time, even if your wiki let's say English, Swedish or
> German...

Yes, I agree. I just wanted to clarify that uninitiated users should not 
choose UTF8-binary on install based on this email thread ;-)

>
> > And choosing it
> > currently is a sure way to get these problems in any case. So to all you
> > users who read this message later: unless you are comfortable with
> > compiling your own DB servers as well, please be careful with such
> > settings ;-) The default settings should normally work.
>
> I really wasn't comfortable with comiliing mysql myself, but after
> some tough time with MW I've found it's the best way to do. But of
> course be warned - configure has lots of options, some should be used,
> some disabled, compile time is long, and even make install is not
> enough - you have to run few scripts - to setup mysql "master" tables,
> to setup init.d and so on...

Sure, again my point was mainly to not have too unexperienced people trying 
things that they cannot handle when finding this mail online. I am sure that 
manual compiling it is the right thing in various situations, but I do not 
wish to answer support requests for cases where such custom setups do not 
work as well as in your case ;-)

-- Markus



-- 
Markus Krötzsch
Institut AIFB, Universät Karlsruhe (TH), 76128 Karlsruhe
phone +49 (0)721 608 7362        fax +49 (0)721 608 5998
[EMAIL PROTECTED]        www  http://korrekt.org

Attachment: signature.asc
Description: This is a digitally signed message part.

-------------------------------------------------------------------------
SF.Net email is sponsored by:
Check out the new SourceForge.net Marketplace.
It's the best place to buy or sell services
for just about anything Open Source.
http://ad.doubleclick.net/clk;164216239;13503038;w?http://sf.net/marketplace
_______________________________________________
Semediawiki-devel mailing list
Semediawiki-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/semediawiki-devel

Reply via email to