paul.harrison at manchester.ac.uk
Thu Jul 2 12:48:01 PDT 2009
During the discussion that took place on this before the latest
Standards document the consensus was that for a document name of the
* That the 2.0 was the version number of the "protocol/standard"
* yyyymmdd was effectively the version number of the document
What was not explicitly thought about is the progression from a
version 1.0 to a version 2.0 of a protocol, - in this case the
document numbering can only start and stay with 2.0 from first WD to
REC - even though there might be substantial (protocol/interface)
changes between the drafts (which would make the programmer in us want
to change the version number). Only after reaching the REC status does
the "2.0" number have to change. What should be removed from the
Standards document is the idea of a public WD that has a 0.x number,
as this is logically inconsistent with the new scheme - the first
public WD of a new protocol/standard should have a 1.0 number - the
new numbering scheme was intended to remove this ambiguity, but I
agree that the Standards document still does not make it clear.
Under this numbering scheme, what is currently VOTable 1.2 should
really be VOTable 2.0 in my opinion, as it causes VOTable 1.1 xml to
be invalid, even though the changes in the written document look quite
minimal. However, I am not sure who will police this sort of issue.....
On 2009-07 -02, at 19:07, Matthew Graham wrote:
> Hi Doug,
> The standards documents says:
> "The number to the left of the (first) decimal point starts with 0
> for documents that are being discussed within a Working Group prior
> to publication for IVOA-wide review. The number increments to 1 for
> the first public version, and to 2, 3, ..., for subsequent versions
> that are not backward compatible and/or require substantial
> revisions to implementations."
> The definition of a Working Draft is:
> "A Working Draft is published at the discretion of a Working Group
> once the WG is satisfied that the document is sufficiently developed
> to merit broader exposure and feedback"
> So I think that a Working Draft fulfils the "IVOA-wide review"
> If the integer progression is only a prerequisite for REC versions
> then the standards document needs to be amended to say this.
> On Jul 2, 2009, at 10:52 AM, Doug Tody wrote:
>> If this were true I agree it would be crazy. But is a WD or PR a
>> "standard"? I should think this would only apply to established
>> standards that are already deployed. That is, to recommendations.
>> Perhaps the documents and standards folks should clarify.
>> Also, I don't think this is just limited to GWS; any DAL or DM
>> for example might also have to deal with this issue.
>> - Doug
>> On Thu, 2 Jul 2009, Matthew Graham wrote:
>>> You might be aware that the Document Standards v1.2 PR has just
>>> completed its RFC (http://www.ivoa.net/cgi-bin/twiki/bin/view/IVOA/DocStdRFC2
>>> ). I raised concerns about the proposed versioning scheme:
>>> "The document now states that there is an integer increment in the
>>> version number in the case where subsequent versions are not
>>> backward compatible. "
>>> and the response that has been posted is:
>>> "The Committee agrees that the version numbering scheme is
>>> challenging when dealing with namespaces and WSDL, and leads to
>>> the conclusion that when IVOA standards describe web services or
>>> have associated XML schemas, with namespaces that when changed,
>>> cause software to break, then these changes must both be
>>> accompanied by an increment to the integer part of the document
>>> and the associated "supplementary" files. This would not affect
>>> most of the standards documents, and should not present any real
>>> logistical difficulty, as there are a sufficient number of
>>> integers available to support any number of revisions. "
>>> This has greatest impact for this working group (GWS but I am also
>>> cross-posting to Semantics) and essentially means that ALL (WD,
>>> PR, etc) versions of our specs with WSDL/XML/RDF documents
>>> (anything with a namespace) will only carry integer versions.
>>> So, for example, the progression of VOSpace 2.0 would actually
>>> proceed as:
>>> VOSpace 2 (first WD)
>>> VOSpace 3 (second WD)
>>> VOSpace 4 (third WD)
>>> VOSpace 5 (first PR)
>>> VOSpace 6 (second PR)
>>> VOSpace 7 (final PR)
>>> VOSpace 8 (REC)
>>> The next version would then VOSpace 9, etc.
>>> Although this is very much a procedural issue, I just wanted to
>>> flag it so that everyone is aware and happy with it before I
More information about the stdproc