Am Montag, 23. Januar 2012, 23:28:33 schrieben Sie:
Package: gosa
Version: 2.6.11-3
Severity: important
User: debian-...@lists.debian.org
Usertags: debian-edu
I discovered this problem in Debian Edu/Squeeze. When a user with a
non-ASCII character in the third position of the first or last name is
created, the generated username contain a non-valid UTF-8 character.
In Debian Edu, our gosa.conf contain the following rule:
idGenerator={%givenName[3-6]}{%sn[3-6]}.
When using UTF-8, the range [3-6] at the moment do not refer to
characters, but bytes in the UTF-8 string. This caused the user 'Gunnar
Grændal' to get a propsed username with gungr?, where ? was half the
UTF-8 bytestream for æ. I believe it would be better if [3-6] refered
to characters 3-6, not bytes 3-6.
As non-ASCII usernames can cause problem when trying to log in to
different systems, I believe it would be best if usernames are always
converted to ASCII, preferably using the //TRANSLIT rule in glibc.
For the example above, one would then end up with echo gungræ | iconv
-t ASCII//TRANSLIT = gungrae as the proposed username.
While it is discussable why this bug is important, 2.7.x packages do the
conversation in another way:
$value = iconv('UTF-8', 'US-ASCII//TRANSLIT', $value);
Same for the 2.6.x maintainance series. You can simply take the function
gen_uids from here
https://oss.gonicus.de/labs/gosa/browser/tags/2.6.13/gosa-
core/include/functions.inc
in order to create a patch for squeeze.
HTH,
Cajus
signature.asc
Description: This is a digitally signed message part.
