[Uni-verse] Fixing a Verse version for the Uni-Verse project

[Sascha Schneider]

Hi all,

At the end of the last Online PCC Meeting we started a discussion about the
fixing of a Verse version for continuing the work in the project. Based on
this version a full and comprehensive documentation (including all function
names, all parameters, all parameter types and their correct order) should
be available. If there are bugs detected within that version (let´s call it
the "Uni-Verse" version of Verse) corresponding bug fixes must be assured.
The fixing of a certain Verse version is essential for the continuous use of
all the tools which are build in the project (e.g. the audio rendering or
the alternative Verse server on which the final use cases will be
demonstrated at the end of the project). As there is no more effort to
adjust the code over and over again to new upcoming Verse versions, this is
important. As we´ve already experienced, the effort of migrating our
applications from one Verse version to another is not negligible. Some
reasons for that are several minor but important lacks in the documentation,
mostly changed function parameter types, -names and -orders for example
which are sometimes hard to find.

However, as there is a work package for the development of the protocol
which will be used in the Uni-Verse project (WP3), it´s a good point to
continue on. According to schedule, this WP was already closed in month 12
so we are on a good way if we try to keep this finished version. So all that
has to be assured is the above mentioned complete and correct documentation
and eventually necessary bug fixes if errors are discovered.

Please don´t get the wrong idea. Of course this argument does not forbid to
continue the development of the Verse protocol and its extension with new
functions or tools. However this should happen outside the Uni-Verse project
so that it is guaranteed, that every tool developed within the Uni-Verse
project supports the same Verse version.

This "Uni-Verse" Verse version is not necessarily the same as the "always
new most currently available" version because it is not possible to continue
the migration of all developed code from one version to the next over and
over during the whole lifetime of the project. At least some years after the
Uni-Verse project is finished the developed software will no longer support
the most current protocol version of course anyway.

So we think, it would be very good for the project and the developing
partners to have a certain Verse version fixed (e.g. R5) to work with in the
the Uni-Verse project.

Recapitulating there are several reasons for that decision:

- First of all the Verse-developing people can concentrate on fixing
remaining bugs in Verse. This would make the protocol even saver, more
stable and reduces their effort for that.
- The next point is, that it takes all Verse-using developers hours or days
to adapt their applications to new upcoming Verse versions that have changes
to the interface. Even if there are only small changes - due to the lacking
of a correct and up-to-date documentation or change notes, it always takes a
lot more time to find all the changed parameters and functionalities.
- This brings up the next point: Having one agreed-upon Verse version for
Uni-Verse would finally allow to write a functional documentation containing
all Verse commands with ALL parameters in the correct order and with the
correct defines & types.
- By doing so it could for example be spent more time on issues like a
threadsafe Verse version (which we think is essential if you want to
establish a serious application for e.g. the industry or the game market) or
fasten up the network transmissions as much as possible.
- Finally it is guaranteed that the corresponding WP is finished and
completed as promised in the project proposal

This is our point of view on that topic. We bring this up at the mailing
list to hear comments from the other partners. In the end, the entire
consortium must decide, but the statements from all partners could be a
basis.

So, what do you think about this proposal?

Regards
Sascha Schneider - Dipl.-Math. - Sascha.Schneider at igd.fraunhofer.de
Tel.: 06151/155631 - Fraunhofer IGD - Fraunhoferstraße 5 - 64283 Darmstadt