4 reasons why ver. # suffice + 3 reasons why more is harmful

[Markus Dolensky]

Hi, Ray,

to my believe the main source of disagreement on your proposed version scheme 
is that we have a different idea of its scope.

The DC needs to enumerate documents - again, documents only - in a way that is 
enough for readers to identify them. This is already the case by assigning a 
release date and title. Even the simplest version number is already redundant 
information in this context.

All the rest of the discussion belongs to software development and maintenance 
- it's nothing that the reader of standard documents needs to worry about and 
therefore such internal bookkeeping doesn't need to be exposed on the official 
doc. tree and hence DC doesn't need to worry about it.


 > So, if can suffer through more of my opinion, read my detailed reactions
 > below.
ditto ...

In the note (http://www.ivoa.net/Documents/latest/DocStdProc.html) it says that 
there can be supplementary resources (like Schema) which the DC publishes as 
is, i.e., embedded version info stays untouched. This will assure consistency 
with the source of information which is most likely some CVS system or some 
other dev. suite that is assigning revision numbers. Mapping such numbers to a 
uniform IVOA scheme implies that there can be 2(!) version numbers for the same 
item.

If you think this is really essential and justifies the overhead of such an 
endless source of potential confusion then I'm not against introducing such a 
scheme: BUT, the mapping of the two namespaces has to be maintained internally 
(for instance on the Wiki) and by the relevant people in each WG and NOT at all 
by the DC. I vehemently object to offload the task of enumerating configuration 
files onto the shoulder of the DC. (For instance, the Wiki uses RCS to 
automatically assign revision numbers to topics and each attachment. It'll be 
fun to maintain a mapping with this kind of source.)


Conclusion:
1. introduce an IVOA version scheme (on top of RCS, CVS etc.) where the 
community considers it essential, but not for the official doc. tree.

2. If it appears inconsistent to have a doc. tree with a different scheme, then 
let's remove version numbers from official documents altogether and simply use 
title and release date to identify them.

Markus