PQL: inline upload will NOT be stored

Douglas Tody dtody at nrao.edu
Fri Jul 10 15:16:39 PDT 2009


On Fri, 10 Jul 2009, Roy Williams wrote:

> I have looked through the PQL specification, and I agree with many of the 
> comments already made.
>
> The one thing that disturbs me most is the idea the uploaded tables are "the 
> same" as other tables, since this makes no distinction between tables that 
> never get stored and those that get stored for for a long time.
>
> What is NOT WANTED for the multicone query is that uploaded tables are stored 
> persistently, since this would imply that an authentication and account is 
> needed, meaning that mashup services and casual users will not be able to use 
> multicone in the simple way (request-response) that we had hoped.

Neither inline nor uploaded (via a URL) tables are persisted
(although I suppose a service could transparently cache a table
referenced via a URL).  Authentication will only be an issue for
tables stored in a per-user VOSpace, once TAP has that capability.
The inline and upload techniques do not require authentication.
In any case, which technique to use for the multi-position table
would be up to the client and user.  We can't prevent some service
implementors from requiring authentication if they want to do so,
but it will not be required for typical queries.

This issue of table uploads is a capability of the TAP engine which
the param query merely uses.  However if we decide to keep the brief
summary of the TAP context then we could add something in there to
clarify this.

The main point about inline/uploaded tables being "the same" as other
tables is that they are available as full-fledged DBMS tables and can
be referenced in any query where any other table might be used.  In
particular an inline or upload table could be used with either an ADQL
or PQL query.

 	- Doug


> I think the PQL spec should say somewhere that:
> "In-line uploaded tables will NOT be stored on the server. Users wishing 
> storage of tables should upload these separately."
>
> Roy
>
> -- 
>
> California Institute of Technology
> 626 395 3670
>



More information about the dal mailing list