Re: RFC: schema locations, transformations [message #1026 is a reply to message #998] |
Mon, 19 April 2004 17:01 |
Nils Poldrack
Messages: 14 Registered: April 2004
|
Junior Member |
|
|
Hello,
I refer to posting of Joachim Buechse 2003-09-25, 18:36
Joachim Buechse schrieb:
> [2] I suggest, that we define the url pattern(s) for RailML schema(s).
> My suggestion is to use the schema version as part of the name. (In the
> currently published examples the namespace definitions refer to files
> rather than urls). I additionally suggest to use separate (sub-)schemas
> for infrastructure, timetable and rollingstock. This allows the usage of
> separate namespaces for the tags. [It will also get more important once
> we contribute the operational data schema drafts which might otherwise
> cloak the schedule for railml 1.0, 1.1, etc].
>
> http://www.railml.org/schemas/<version>/<name-without-xsd>
>
> http://www.railml.org/1.00/railml
> http://www.railml.org/1.00/infrastructure
> http://www.railml.org/1.00/timetable
> http://www.railml.org/1.00/rolingstock
>
There was no comment to this proposal. But now I suggest to define the
URL patterns the other way round as Joachim proposed. My favourite URL
pattern is
http://www.railml.org/<scheme>/<version>
Reasons for this rotation:
1. There is more order in this system.
2. We do not have "empty" directories with only one scheme. (This might
appear when there is only a v1.05 from timetable, but even a 2.x from
all other schemes.)
3. It is immediately visible with is the newest version of a scheme.
I'm just creating a pattern document for the documentation of changing a
railML-scheme. There MUST be written down the storage of all documents
and schemes. And a uniform path is REQUIRED.
Thank you for your comments,
best regards
Nils P.
PS. If there is no comment or veto against, I will write as proposed.
|
|
|