TAP Implementation Notes

[Mark Taylor]

One other item.

On Mon, 8 Oct 2012, Markus Demleitner wrote:

> (3) Are there any other quirks you'd like to fix in the next round of
>     standardization?
> (4) Are there any other features you'd like to see in the next round
>     of standardization?

Currently the TAP standard says that both synchronous and asynchronous
modes are mandatory for a TAP endpoint.  I know of at least two
TAP services which simply don't implement the asynchronous mode,
because any query they are expecting to service will be small
enough that async operation is not worthwhile.  They therefore
don't count as compliant TAP services.  Given the 
considerable additional effort involved in implementing and
operating an async service, one can see their point.
The mandatory nature of asynchronous operation therefore seems
to be in *some* cases expensive and unhelpful from a service
point of view (though of course the utility of allowing async
where appropriate is not in question).

>From a client point of view, there is some benefit in having a single
mode that can always be used, and async could be it, but in practice
if you can behave as an async client it is probably easy to implement
synchronous operation, and where the service only wants to provide
synchronous operation it's likely to be more efficient and less
error-prone all round to do it that way than to go async.

So: what about in a future revision of the TAP standard making
the async mode optional rather than mandatory, while leaving
sync mode mandatory?  TAPRegExt would need a corresponding edit
to allow declaration of supported modes.

I have a vague impression that this might have been the subject of
one of the great TAP Wars in which I was not intimately involved,
so if it's too painful to reopen, pretend I didn't say it.

Mark

--
Mark Taylor   Astronomical Programmer   Physics, Bristol University, UK
m.b.taylor at bris.ac.uk +44-117-9288776  http://www.star.bris.ac.uk/~mbt/