[tex-live] User names too longs or with diacritics on Windows

[Zdenek Wagner]

2013/4/13 Philip TAYLOR <P.Taylor at rhul.ac.uk>:
>
>
> Denis Bitouzé wrote:
>
>> Le vendredi 12/04/13 à 22h29, karl at freefriends.org (Karl Berry) a
>> écrit :
>
>>> What should it return?  Looks like all kpse is doing is returning
>>> the directory as given ...
>
> Well, is it ?  My test case :
>
> Test file : D:\Users\Cécile\Cécile.txt
> Current directory : D:\Users
> Command: kpsewhich --path=.// Cécile.txt
> Results returned : ./CÚcile/CÚcile.txt
>
> where "Ú" in this message is "Ɵ" in the CMD window.
>
> On what basis can it be asserted that "kpse is returning
> the directory as given" ?
>
There are IMHO two design bugs in Windows:

1. Automatic generation of short names. It is near to impossible to
predict whether the suffix will be ~1, ~2, ... From the programmer's
point of view (I am not a Windows programmer so that I cannot give
precise details) there are two APIs, one is able to use the real long
name, another can only use the short name

2. Usage of different encodings in GUI and cmd. This stems from MS
DOS. For instance, in Czech locale CP852 is used in cmd (as in the old
MS DOS) but CP1250 in GUI. If you type Czech document in Notepad, it
will be in CP1250. Nowadays several Windows text editor can use even
UTF8. If you open cmd prompt and display it by "type", it will be
wrong because CP852 is expected. I suspect that French Windows expect
CP850 in the cmd prompt and something else in GUI and kpsea library.

> Philip Taylor
>



--
Zdeněk Wagner
http://hroch486.icpf.cas.cz/wagner/
http://icebearsoft.euweb.cz