[postgis-users] Re: Re: Proposed SQL interface for PGRaster

[Patrick]

"Mark Cave-Ayland" <mark.cave-ayland at ilande.co.uk> wrote in message 
news:1165409782.5590.40.camel at mca-desktop...
> On Wed, 2006-12-06 at 12:58 +0100, Patrick wrote:
>
>> Hi Mark,
>>
>> Endianness of the data is of no concern on the DB side. It only becomes 
>> an
>> issue on the application side. I would suggest to leave the storage 
>> layout
>> up to the application creating the raster and have an attribute 
>> indicating
>> the endianness. This implies that the application will have to do any
>> required conversion, but I expect many data-user applications to be on 
>> the
>> same platform as data-producer applications.
>>
>> Cheers,
>> Patrick
>
>
> Hi Patrick,
>
> Are you sure? What about the proposed raster to raster transformations?
>
Hi Mark,

Well, you got me there! However, as you wrote in another post in this thread 
"A quick scan of the GDAL API hints that GDAL itself converts between 
file/native data types when returning raster bands as C arrays." This IMHO 
is true of all client-side applications that can access a variety of 
formats, so I see the endianness not as an issue, but as a property of the 
data. All processing code, client-side and server-side, should be able to 
convert endian order, rather than impose a restriction on the design.

Cheers,
Patrick