Your message dated Sun, 10 Aug 2014 11:36:44 +0200
with message-id <op.xkdebic9lhaakn@saturne>
has caused the report #756386,
regarding dblatex: Wrong indexterm sorting order for Norwegian Bokmål (nb)
to be marked as having been forwarded to the upstream software
author(s)
(NB: If you are a system administrator and have no idea what this
message is talking about, this may indicate a serious mail system
misconfiguration somewhere. Please contact [email protected]
immediately.)
--
756386: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=756386
Debian Bug Tracking System
Contact [email protected] with problems
--- Begin Message ---
Hello,
By itself the backend has no side effect on the index building, and
therefore selecting the backend from language will not help. In fact the
behaviour is different with XeTeX because makeindex fails to build a valid
index with headings with UTF8 characters, and therefore dblatex retry to
build an index by removing the style with headings (for Andreas, it is
located here: lib/dbtexmf/dblatex/grubber/index.py line 166). But in any
case makeindex cannot sort characters beyond latin charset, which explains
that it puts everything it doesn't know in a symbol heading, or in a
separate unsorted list when it fails with headings.
As a workaround if it can be usefull, a post-process script can be used to
force makeindex to sort words correctly, by transforming indexes like this
example: \index{Ærlig talt} => \index{ae@Ærlig talt}. The characters
before the '@' are used as sort keys. The limitation is that it will not
put the word in a separate heading 'Æ' but in the heading 'A'.
So, the default makeindex tool is not suited to sort UTF-8 chars, and
should be replaced by xindy which is aware about the language used and
UTF8 chars. Dblatex should be able to handle xindy but I haven't give a
try yet. I'll try to make it work when I have an internet bandwidth to
install it (i'm on hollidays).
Regards,
BG
On Tue, 05 Aug 2014 20:10:00 +0200, Andreas Hoenen
<[email protected]> wrote:
Petter Reinholdtsen <[email protected]> wrote:
[Andreas Hoenen]
> Hi Petter,
Hi. :)
> This looks like a TeX problem to me: when handled with the default
pdftex
> backend, the indexterm sorting order is wrong, however when switching
to
> the xetex backend, it's correct. XeTeX just seems to be better suited
> for many non english languages than traditional LaTeX.
Could be. But should not '<book lang="nb">' tell dblatex to do the
right thing for Norwegian Bokmål? :)
One could think about making the default backend language dependent.
@Benoît: putting you into the loop: what do you think about this idea?
> Thus you should give the xetex backend a try.
I did, using 'dblatex -b xetex sort-order-test.xml', and the new
index end up with this ordering instead:
Alle inne
Bare tull
Åndalsnes
Ærlig talt
Øvrigheten
The 'Symbols' header is gone, as there is no symbols/letter headers in
this version. Still wrong ordering, though, but different. It should
be
Alle inne
Bare tull
Ærlig talt
Øvrigheten
Åndalsnes
Hm, this would be a XeTeX problem, dblatex just emits a \printindex
command.
Regards, Andreas
--- End Message ---
