[postgis-devel] [raster] ST_Binary semantic

[Mateusz Loskot]

On 28 February 2013 16:30, Bborie Park <dustymugs at gmail.com> wrote:
>>
>> Turns out it is an important issue, ticket reported:
>>
>> http://trac.osgeo.org/postgis/ticket/2217
>>
>> Best regards,
>> --
>> Mateusz Loskot, http://mateusz.loskot.net
>
> This one is strange.  Should ST_AsBinary be returning a faithful
> representation of the raster in the database, which would include the
> out-db path?  Or should ST_AsBinary be returning a complete -- and
> different -- representation, including the out-db data instead of the
> path?


As Sandro said on IRC:
"canonical representation of WKB doesn't necessarily have to match
ST_AsBinary output"
and IMO, in this case it shouldn't.

> I've read through the IRC logs and am fine with adding a new function
> so that ST_AsBinary's present behavior persists.  Any ideas what this
> function would be called?  This new function will be needed in GDAL
> when it supports out-db rasters (it currently does not).

Can we consider changing ST_Binary behaviour to make it consistent
regardless storage mode?

Four people on #postgis support such consistency as important factor,
you've seen it itself.

Regina mentioned danger of broken backward compatibility,
but at the same time agreed that at this still early stage of PostGIS Raster
adoption, we still have time to correct such mistakes.

Best regards,
--
Mateusz Loskot, http://mateusz.loskot.net