Home » railML newsgroups » railml.misc » Concerns about RailML interoperability
Concerns about RailML interoperability [message #1100] Fri, 06 May 2011 11:56 Go to next message
tuomas.tiihonen is currently offline  tuomas.tiihonen
Messages: 15
Registered: May 2011
Junior Member
Hi,

I am concerned. My concern is the semantic interoperability of the RailML
standard now and in the future.

RailML syntax is defined in schema files, so it is easy for any member of
community or outside of community to make implementation that can accept
RailML as long as it passes XML validation. Problem comes from semantics.
As I have understood, the semantics are documented only in wiki pages.
Wiki pages are not complete and creates this huge risk for fragmenting the
railML to different branches, where each branch is one-to-one agreement
between users of the railML.

How is this handled currently? Does every implementation using railML have
separate agreements of semantics? Are there n-amount of undocumented
semantics?

Does the community recognize this problem and are there any actions
ongoing to make the semantics complete? I have now only power of guessing
in most parts of railML semantics.

I know that there exists standards where this problem has risen too late.
So late that there can be tens or hundreds of different implementations of
valid standard-using-software, but that can only inter operate
out-of-the-box with 1-3 other implementations.

This is huge stumbling stone and must be overcome to keep railML usable.
Overcoming comes more difficult day by day.

With Kindest Regards,
Tuomas Tiihonen

--
----== posted via PHP Headliner ==----
Re: Concerns about RailML interoperability [message #1124 is a reply to message #1100] Tue, 06 November 2012 10:14 Go to previous message
Susanne Wunsch is currently offline  Susanne Wunsch
Messages: 180
Registered: March 2008
Senior Member
Hello Tuomas and others

I'm sorry for the late answering of this posting. In that time I was not
aware of our quite good documentation in the wiki pages.

tuomastiihonen(at)mitroncom (Tuomas Tiihonen) writes:

> I am concerned. My concern is the semantic interoperability of the RailML
> standard now and in the future.

I know that other users share this concern with you. Some want a
_stable_ railML. But if new customer wishes arise they want a _flexible_
railML. ;-)

> RailML syntax is defined in schema files, so it is easy for any member of
> community or outside of community to make implementation that can accept
> RailML as long as it passes XML validation. Problem comes from semantics.
> As I have understood, the semantics are documented only in wiki pages.
> Wiki pages are not complete and creates this huge risk for fragmenting the
> railML to different branches, where each branch is one-to-one agreement
> between users of the railML.

We try to develop XML Schemas that don't fragment into different
branches. The sub-schema coordinators try hard not to define further
redundancies in order to keep the schemas specific to their
component. We just inherited some redundancies and sub-schema-specific
definitions of the same topic from the very first railML version
1.0. But with the next major release of railML (3.0) we will get rid of
them.

Anyway there will be one-to-one agreements on how to use railML in a
specific use case. Each software tool has a special focus on data. Only
data that can be handled in the software tool may be exported with
railML. On the other side a software tool will only import railML
components it may handle, others will be dropped or kept unused.

> How is this handled currently? Does every implementation using railML have
> separate agreements of semantics? Are there n-amount of undocumented
> semantics?

We hope that the semantics are most clearly documented in the wiki pages
as possible. If not, please enhance the wiki pages with your
experiences. We know that there are some wiki pages (especially
regarding the infrastructure sub-schema) that are not satisfactory
filled with semantic content.

The one-to-one agreements should mostly restrict the railML components
to a subset that ist needed for the exchange. Additional extensions that
are not of common interest for railML may also be defined in extension
XML schemas.

But nevertheless the semantics should be clearly stated at the railML
documentation (wiki).

> Does the community recognize this problem and are there any actions
> ongoing to make the semantics complete? I have now only power of guessing
> in most parts of railML semantics.

Please, don't guess. Ask in the forum for each topic! (We try to answer
more quick!)

What else than the wiki documentation, the public forum and the XML
Schemas would you suggest for overcoming the problem of clear defined
semantics?

Thanks for sharing your concerns and experiences with us. I hope we will
manage this problem step by step in a more satisfying way.

Kind regards...
Susanne

--
Susanne Wunsch
Schema Coordinator: railML.common
Previous Topic: A train with ETCS?
Next Topic: Required 'id' attributes (was: More detailed 'speed change' definitions)
Goto Forum:
  


Current Time: Fri Aug 18 05:12:35 CEST 2017