Re: Insert link with "foreign" character - cannot save

[Sebastian Rose <sebastian_rose@gmx.de>]

OK, this is a severe Emacs 23 bug.


I can reproduce the errors like this:


sh$ mkdir test
sh$ cd test
sh$ touch ümläute.org
sh$ ls
ümläute.org
sh$ convmv --notest -f utf-8 -t iso-8859-1 ümläute.org 
Your Perl version has fleas #37757 #49830 
mv "./ümläute.org"	"./�ml�ute.org"
Ready!
sh$ ls
?ml?ute.org


Now I open an Org-file in Emacs 23, and do

  C-u C-c C-l

Navigate to that file and

BAMM - cannot save my work anymore!!!



Another nice test: Try to recode an UTF-8 encoded file name with German
Umlauts unsing

  M-x recode-file-name 

It does not work either.



I'll file an Emacs bug report as this is no bug in Org-mode.




  Sebastian





Sebastian Rose <sebastian_rose@gmx.de> writes:
> Carsten Dominik <carsten.dominik@gmail.com> writes:
>
>> Hi Mattias,
>>
>> I tried that, and my buffer swiched to unicode encoding automatically.
>>
>> Unfortunately I don't know much about coding systems, and so I do
>> not know how to fix this.
>>
>> Anyone????
>>
>> - Carsten
>>
>> On Nov 12, 2009, at 10:28 AM, Mattias Jämting wrote:
>>
>>> (I'm using English Windows Vista x64, Emacs 23.1 and Org-mode 6.32b)
>>>
>>> So i'm doing C-u C-c C-l to browse for a file in order to insert a link to
>>> it.
>>>
>>> The path and/or the filename contains for instance an ö (an o with two dots
>>> above it, also the swedish word for "island"), which gets translated in my
>>> org-file as \366.
>>>
>>> When I try to save the file I see the message:
>>>
>>> These default coding systems were tried to encode text
>>> in the buffer `jwd.org':
>>>   (utf-8-dos (79 . 4194294))
>>> However, each of them encountered characters it couldn't encode:
>>>   utf-8-dos cannot encode these:  These default coding systems were tried
>>> to encode text
>>> in the buffer `jwd.org':
>>>   (utf-8-dos (79 . 4194294))
>>> However, each of them encountered characters it couldn't encode:
>>>   utf-8-dos cannot encode these:  \366
>>>
>>> Next I tried to hack myself a fix :-)
>>>
>>> I added (?\366 . "%F6") to org-link-escape-chars and ran make on it again,
>>> but it didn't seem to work.
>
>
> It's true, you cannot encode the bytes with dec. values above 127 in
> utf-8 (see `man utf-8', unicode.org, whatever).
>
>
>
> Seems your filenames are not utf-8 encoded.
>
> Here it works, because on current Linux distros (and MAC OS??) filenames
> are utf-8 encoded:
>
>   C-u C-c C-f
>   nä TAB RET
>
>   [[file:nächtes-nötiges.org][Umlaute in Dateinamen]]
>
>
> Maybe, if the file is from an old system, rename it (twice, to give it
> the original name) and try again. Would that work?
>
>
>
>
> There's an interesting discussion going on on emacs-devel, that might be
> related (but it's not about filenames). You may read the entire thread
> here:
>
>   http://lists.gnu.org/archive/html/emacs-devel/2009-11/msg00661.html
>
>
> The fazit so far, as I understood it, is, that Emasc 23 distinguishes
> single and multibyte strings. Better not use array functions to handle
> strings (which are multibyte internaly) in Emacs 23.
>
> The OP did
>
>   (setq nl "\n")
>   (aset nl 0 ?ñ
>   (insert nl)
>
> which sets the first _byte_ of an array to 241, which in turn has no
> valid representation on screen as character in Unicode (see `man urf-8'
> and unicode.org). Thus Emacs insert \361 - for some reason :)
>
>
>
>    Sebastian
>
>
> _______________________________________________
> Emacs-orgmode mailing list
> Remember: use `Reply All' to send replies to the list.
> Emacs-orgmode@gnu.org
> http://lists.gnu.org/mailman/listinfo/emacs-orgmode