On Wed, 19 Dec 2007 22:01:14 +0100 Steffen Wolfrum wrote:

Am 19.12.2007 um 14:45 schrieb Wolfgang Schuster:

...

2007/12/19, Steffen Wolfrum :

...

Hi,

the following (with mkii) is probably not the most pressuring topic to most of you ... apologies!

Although I have set \mainlanguage[de] the register seems to be unaffected. All my cultural heritage (i.e. the Umaut) is ignored!!!

o(

The problem is the sorting – keywords containing umlauts seem to be sorted after "e":

Bestimmung Bühler Bildung

Are there language-specific arguments/options for \placeregister[index]?

Thanks,

Steffen

Hi Steffen,

Can you post a example where this did happen for you.

Hi Hans and all others, I found the reason for the wrong index sorting in Steffens files. The correct sorted file contains something like this. \enableregime[utf] \starttext text\index{Bestimmung} text\index{Bühler} text\index{Bildung} text \placeindex \stoptext The text is enclosed by the normal start/stoptext pair and the document setup is written before the start of the document while the wrong file contains this: \starttext \enableregime[utf] text\index{Bestimmung} text\index{Bühler} text\index{Bildung} text \placeindex \stoptext The files did not really contain start/stop but use ConTeXts document structure and \everystarttext is executed by \startpoduct, the first command in the product file. The problem is now, ConTeXt write information for the index sorting into the tui file and a few additional entries for every spcifiec file encoding, in this utf-8. This works quite well in the first example because \enableregime is writte before \starttext while in the second example the sorting information are written before ConTeXt knows the file encoding. A workaround for the moment is to write \enableregime before \startproduct in the main file but I hope this coule be fiyed in the next release. I still wonder why nobody noticed this untill now. Wolfgang

On Fri, 21 Dec 2007 18:34:32 +0100 Hans Hagen wrote:

Wolfgang Schuster wrote:

...

The problem is now, ConTeXt write information for the index sorting into the tui file and a few additional entries for every spcifiec file encoding, in this utf-8.

writing the sort vector is hooked into starttext

I realized this myself, because the only difference between both version had been the file size of the tui files.

...

This works quite well in the first example because \enableregime is writte before \starttext while in the second example the sorting information are written before ConTeXt knows the file encoding.

A workaround for the moment is to write \enableregime before \startproduct in the main file but I hope this coule be fiyed in the next release.

i'll find another hook

...

I still wonder why nobody noticed this untill now.

maybe because of \expanded names, \eacute still sorts somewhat right

I found this solution also but it is easier to write ä, ü ... instead of \aumlaut, \uumlaut etc. BTW, why could I use only \aumlaut in macros but not \adiaresis. Wolfgang

On Fri, 21 Dec 2007 19:35:32 +0100 Hans Hagen wrote:

Wolfgang Schuster wrote:

...

BTW, why could I use only \aumlaut in macros but not \adiaresis.

diae

D’oh! Wolfgang