TAP RFC (inline UPLOAD)

[Doug Tody]

Hi Pat -

I originally suggested this and it would work, but in a later posting
(see http://www.ivoa.net/forum/dal/0909/1492.htm) I suggested a
more explicit connection with the parameter mechanism which we could
consider (there was no subsequent discussion).  The main change would
be to replace "inline:" with "param:".  I won't repeat the rationale
here as it is in the referenced posting above.

Also, if we consider:

     > UPLOAD=table_name,inline:some_table_name

It is not necessary for the referenced datum ("some_table_name" in
this example) to be a table name since we already specify the actual
table name on the left side.  In the parameter-based approach the
uploaded object would be associated with a parameter name assigned by
the client.  Otherwise it is arbitrary so long as any HTML restrictions
are met.  The mechanism is in fact a general file upload mechanism
and is not limited to tables.

If we use "inline:" instead of "param:" it will still work and will
accomplish the same thing, but there is a danger of a name collsion
with a HTTP parameter as Tom originally suggested.

> Can someone implement a crude test to see that it is a workable upload
> solution?

If you upload a file on a TWiki page it is using the same mechanism;
that is where we originally got this from.

 	- Doug



On Wed, 16 Sep 2009, Patrick Dowler wrote:

> On Monday 14 September 2009 02:00:09 Keith Noddle wrote:
>> TAP RFC to 25th September to allow the above details to come together.
>
> on issue #7 (inline table upload) Doug and Tom have proposed a solution as
> follows:
>
> caller would specify the UPLOAD parameter with pair of value, e.g.
>
> UPLOAD=table_name,inline:some_table_name
>
> where "table_name" is the name of the table as used in the query (eg
> TAP_UPLOAD.table_name) and the URI uses a special scheme "inline" to denote
> that the content is inline with name="some_table_name". Thus, "inline" URI
> scheme will be described (in the spec) in the UPLOAD parameter section.
>
> Reminder: one can already use things like:
>
> UPLOAD=mytable,http://examplecom/some_table.xml
>
> or
>
> UPLOAD=mytable,vos://example/some_table.xml
>
>
> Have I got this right? Any objections/problems to this fix?
>
> Can someone implement a crude test to see that it is a workable upload
> solution? before the TCG review? I am not sure when we (CADC) will get to
> implementing upload...
>
>