Hi Keith -
On Thu, 15 May 2008, Keith Noddle wrote:
> I would like to thank Doug for uploading his version to the same page. I
> need to make one small correction to his email: the specification he
> references is the TAP/Param Draft even though it goes some way beyond
> merely specifying parametrised querying (see below).
Well I am surprised to hear you characterize the scope of "TAP/Param" this way! After all we checked with you a month ago before beginning work on the draft spec, to clarify the scope of TAP/Param (see mail below). We agreed that TAP/QL and TAP/Param would likely have similar scope, differing mainly in emphasis (all ADQL in one case vs both ADQL and Param query methods in the other).
If we had not agreed upon this key point then I am not sure what we would have done, but we could never have agreed to merely focus on a non-ADQL, purely parameter based interface. ADQL, VOSpace/async integration, multi-region queries etc. are an essential requirement for NVO as well (and we have been doing this here for 10-20 years or so); it is just that our requirements are evidently broader than those of some other folks. But we could never proceed with prototyping of "TAP/Param" this summer if all it did was ParamQuery. We need one TAP which encompasses all primary requirements in one interface, and this is what is proposed.
> I would like to thank Doug for uploading his version to the same page.
Just to clarify, this is not "my" version. While I did most of the writing of this in the short several weeks available, it is mainly just an elaboration of the agreements we reached in the international TAP tiger team meeting last November. Plus we have had numerous sessions on this within the NVO over the past 6 months at least, and had what additional discussion there was time for in the several weeks since the IVOA teams were formed.
It might be worth reviewing the minutes of the TAP tiger team meeting again, as so far as I know this represents the best attempt thus far to try to get everone on the "same table" (no pun intended), and agree upon the key issues and requirements for TAP. Agreed everyone could not be there of course, and broader discussion is needed, but given that we all flew in from around the world and spent two days discussing all this in detail, making considerable progress discussing details such as sync/async, metadata queries, vospace integration, etc., it is a shame to not make more of an effort to at least take these agreements as a starting point. (Again, see http://www.ivoa.net/cgi-bin/twiki/bin/view/IVOA/TapJhu).
[...]
> [From Doug:]
> Yes, I will be happy to lead the "TAP/Param" effort. The scope of what
> we are doing however, is what we discussed in the November tiger team
> meeting, including both AdqlQuery and ParamQuery operations within
> the same interface, as this is the only approach which meets all
> our requirements. Initial prototyping will focus more on ParamQuery
> though. We are still working on this, but should have at least an
> initial spec and some prototyping to discuss at the interop.
Excellent, that is good news. I'm hoping the same will be true for TAP/QL with the emphasis on ADQL. If so, we will have things of substance to discuss - and maybe the edges will blur between TAPs Param and QL after all... Received on 2008-05-15Z22:47:55