|
Home > Archive > Unix administration > September 2005 > file name and hostname in UNICODE
You are viewing an archived Text-only version of the thread.
To view this thread in it's original format and/or if you want to reply to
this thread please [click here]
| Author |
file name and hostname in UNICODE
|
|
| allab@sympatico.ca 2005-09-28, 6:02 pm |
| Hello guys,
We are modifying our application to accomodate UNICODE. Our application
uses TCP/IP and file resources a lot.
Please let us know if UNIX file names and hostnames can be in languages
other than English and if they can be in UNICODE.
Are there other UNIX system resources that we should keep in mind while
making our app work with UNICODE?
Please reply to allab@sympatico.ca
Thank you,
Alona
| |
| Michael Vilain 2005-09-29, 2:55 am |
| In article <1127937945.836788.270330@g44g2000cwa.googlegroups.com>,
allab@sympatico.ca wrote:
> Hello guys,
>
> We are modifying our application to accomodate UNICODE. Our application
> uses TCP/IP and file resources a lot.
>
> Please let us know if UNIX file names and hostnames can be in languages
> other than English and if they can be in UNICODE.
>
> Are there other UNIX system resources that we should keep in mind while
> making our app work with UNICODE?
>
> Please reply to allab@sympatico.ca
>
>
> Thank you,
> Alona
I would seriously doubt it. The standard for many years has been ASCII
characters excluding certain characters. You might be able to name a
machine something in UNICODE, but I don't think DNS or NIS or LDAP is
UNICODE compatible so what would be your point?
Better stay within the standard UNIX namespace of ASCII letters, first
must be a letter, and no punctuation.
If this is a requirement, what systems will it interoperate with?
Windows, UNIX, or something else? If you're in a stand-alone
environment, you might get away with it, but the rest of the internet
may not like it.
--
DeeDee, don't press that button! DeeDee! NO! Dee...
| |
| Doug Freyburger 2005-09-29, 6:00 pm |
| allab@sympatico.ca wrote:
>
> We are modifying our application to accomodate UNICODE. Our application
> uses TCP/IP and file resources a lot.
>
> Please let us know if UNIX file names and hostnames can be in languages
> other than English and if they can be in UNICODE.
File names do not have any specific standards other
than forbidding forward slash and null, but the
fact that you can create files with unicode name
does not mean you should.
Hostnames have a more powerfull issue. DNS requires
simple character names in A records but has no such
requirement in CNAME records. It is possible to
have names as simple as serial numbers for the A
records and use CNAME records with unicode. Of
course this sort of stunt is an abysmally bad idea:
Every time the host name is displayed it uses the
A record or PTR record and both must use the simple
set. Doing it would confuse the users.
Languages other than English does not have to be
a unicode issue, though. Ich kann auf Deutsch
ohne unicode screiben. Most/all languages should
offer a transliteration method that is all
desirable than unicode but that still works.
> Are there other UNIX system resources that we should keep in mind while
> making our app work with UNICODE?
This is the "should" issue I mentioned with filenames.
If you end up with shell special characters, or if
even one of the applications you use is not unicode
compatible you're in trouble.
So try it with files in a test environment and see
if all of your programs handle it. Forget hostnames.
Do them phonetically in other languages if you wish
but stay with simple characters. En francais, non
sedilla.
| |
| allab@sympatico.ca 2005-09-29, 6:00 pm |
| Doug,
We are not going to name our computers or files in other languages. We
developed an application that must read computer names and file names.
So far, our application has been non-UNICODE and our clients around the
world were required to use English with our program.
Now we want to be more user-friendly and we want our program to "speak"
other languages. Can our UNIX clients in Italy or Germany have a
computer name that contains national characters? Can our Russian
clients have files containing Russian national characters? We are not
talking about transliteration, but we want to accomodate real life
situation.
Regards,
Alona
| |
| Doug Freyburger 2005-09-29, 6:00 pm |
| allab@sympatico.ca wrote:
>
> We are not going to name our computers or files in other languages. We
> developed an application that must read computer names and file names.
> So far, our application has been non-UNICODE and our clients around the
> world were required to use English with our program.
>
> Now we want to be more user-friendly and we want our program to "speak"
> other languages. Can our UNIX clients in Italy or Germany have a
> computer name that contains national characters?
No. The RFCs require that hostnames use a very limited
set of characters.
> Can our Russian
> clients have files containing Russian national characters?
Yes. UNIX filenames are not restricted by those RFCs.
> We are not
> talking about transliteration, but we want to accomodate real life
> situation.
It would be a very good idea to translate all of your
messages and canned responses by language. Allow your
clients to use what filenames they wish. That choice
is not available with hostnames. Two different
situations, two different answers.
| |
| Villy Kruse 2005-09-30, 2:50 am |
| On 29 Sep 2005 09:02:27 -0700,
Doug Freyburger <dfreybur@yahoo.com> wrote:
> allab@sympatico.ca wrote:
>
> File names do not have any specific standards other
> than forbidding forward slash and null, but the
> fact that you can create files with unicode name
> does not mean you should.
>
When talking about UNICODE you also should think about its encoding.
For utf-8 encoded UNICODE the kernel code can hardly tell the difference
between that and any other 8-bit encoded character set. Utf-16, however,
when seen as a string of 8-bit values, usualy contains too many binary
zeroes to be valied file names as far as the kernel is concerned.
Villy
|
|
|
|
|