new TAP-1.1 WD released

[Patrick Dowler]

I guess I was a little overzealous when writing that section. What I
was thinking is that the datatype="char" size="32" (eg) one would use
in a TAP-1.1 service cannot convey the variable length, which is why I
added "or variable-length value".... I guess it would be OK if such a
thing was described as datatype="char" size="32" arraysize="32*" (in
the tap_schema and votable) and the variable-ness is only conveyed in
the arraysize. [will fix]

Pat

On 2 August 2017 at 05:41, Grégory Mantelet
<gmantele at ari.uni-heidelberg.de> wrote:
> Hi DAL,
>
> About the TAP-1.1 WD (2017-07-07), I think there is a mistake about the
> description of the deprecated column "size" (of TAP_SCHEMA.columns).
>
> In TAP 1.0:
>
> The “size” gives the length of variable length datatypes, for example
> varchar(256); this size does not map to the VOTable arraysize attribute when
> the
> latter specifies the size and shape of a multi-dimensional array.
>
> But in this TAP 1.1 WD:
>
> The "size" column is retained for backwards compatibil-
> ity to TAP-1.0 and must contain the integer value equivalent to arraysize
> when possible and must be null if arraysize represents a multi-dimensional
> or variable-length value.
>
> I totally agree about the fact that "size" should be NULL when the column
> values are multi-dimensional arrays.
> But I do not agree that "size" no longer represents the length of a
> variable-length values (i.e. varchar, varbinary, char and binary).
> It does fit with the TAP 1.0 description anymore. It may also confuse
> users/clients when reading old (i.e. TAP-1.0) and new (i.e. TAP-1.1)
> services.
>
> Is it normal? Or should it be fixed?
>
> Cheers,
> Grégory



-- 
Patrick Dowler
Canadian Astronomy Data Centre
Victoria, BC, Canada