| coding of dc:language [message #1109 is a reply to message #1084] |
Fri, 30 March 2012 19:53   |
Dirk Bräuer
Messages: 311
Registered: August 2008
|
Senior Member |
|
|
Hallo,
I have added some examples on using Dublin Core Metadata Set to the Wiki
pages. There was one response from Susanne concerning the item
<dc:language>.
Generally, this item shall be used to code character set of the names (e..
g. station names a. s. o.) of the RailML file. This value is of importance
in case the containing Unicode names have to be converted into a
non-Unicode-string by the reading software.
Originally, I wanted it to contain the Codepage Number of the data
<dc:language>1252</dc:language> ;(1252=ANSI - Lateinisch I)
because, from my experience, one does need a Codepage Number do convert
non-Unicode strings.
This did not enjoy Susanne who rather would prefer a coding like ISO 15924:
<dc:language>de-CH</dc:language>
The problem is that there is no 'conversion table' or something like that
(as far as I know) to convert Codepage Numbers into ISO 15924 codes or
vice versa. There is, unfortunately, no standardisation of Codepages at
all. So if we do not allow the non-standardised Codepage Numbers we cannot
tell the reading software how to convert the UTF-8 strings of a RailML
file into non-Unicode strings. This leaves a reading software with the
need to 'scan' the names for special characters and deduce a Codepage from
this - a more empiric solution.
The problem with the ISO 15924 codes is not only that there is no
'conversion table'. It is also that typically a RailML file contains names
of more than one language, e. g. some foreign station names also. This is
normally no problem because one Codepage normally allows languages of
neighboring countries. Our 'middle-European' Codepage (1250) allows German
Umlauts, Czech 'háčeks' (carons) and Sorbian/Polish 'striked-out L's'. But
what should we write into <dc:language> if a RailML file contains all of
these three and the writing programme only know that it is CodePage 1250?
Anyway, it is not a big problem because it only applies to non-Unicode
software and there should be not much non-Unicode software nowadays. It's
only that we do not know...
So, from my opinion we have two possible solutions:
a) either to skip this <dc:language> at all and delete it from all
examples
b) or still to allow and recommend a Codepage number (!) there because it
costs nothing, may help someone, and there is no other need for this
element.
It does not make sense to code it with ISO 15924 since, as I did explain,
there is normally not _one_ source language for all the RailML file.
@Susanne: If nobody answers this 'post' in a near future you can tell me
at any time to delete this <dc:language> from the examples without further
objection from me. I leave it up to you, doing nothing more from my side..
With best regards,
Dirk.
|
|
|
|
Members
Search
Help
Login
Home
