rplante at poplar.ncsa.uiuc.edu
Thu Apr 1 04:24:59 PST 2004
On Thu, 1 Apr 2004, Marco C. Leoni wrote:
> - do we really need 10000 versions in between ver. 0.1 and the first
> published WD (i.e. ver. 1.0)?
No, of course not. But I expect that it won't be uncommon that we'll
need more than ten.
> - the documents process does not take into account the xml files et
> similia: they are "supplementary resources" that means you can follow
> your preferred numbering schema (CVS, RCS, VSS, etc.).
In some cases where a document has an associated standard schema (e.g.
ADQL), it would advantageous to have the XML version track the document
version. Even in general, people will want to just adopt one system as a
matter of practice.
Consider the situation of VOResource, whose version now sits at 0.9. We
have a choice between whether to make a major fundemental change to the
core model or just an incremental change. (This is not hypothetical.)
In my suggested system, it's a choice between going to 0.10 or 0.9.1. In
the IVOA system, I'm pretty much forced to go to 0.91.
Is there functional disadvantage to my system?
(I hope I'm not overblowing this one.)
> Ray Plante wrote:
> >Hi Bob, Markus, Marco,
> >I know I should have said something earlier, so feel free to say, "it's
> >too late to change it."
> >Thanks for the acknowledgement in the Guidelines Note regarding version
> >numbers; however, you modified my suggestion in toward a direction I was
> >trying to get away from. The convention of having 0.21 come between 0.2
> >and 0.22 has a real problem: you only get ten revisions of a particular
> >level. What comes after 0.99 if you are not ready for 1.0? 0.991? Or
> >are you out of luck? The result is that you no longer have a sense of how
> >major a change the revision is.
> >The point of *my* saying "fields on either side of the period (.) are
> >integers" is to say that 3 comes between 2 and 4, and 21 come between 20
> >and 22. All increments between a set of periods are considered the same
> >level of change.
> >This is how RCS, CVS, and probably most other revision control systems
More information about the stdproc