[postgis-devel] [PostGIS] #2122: [raster] Real extent feature lost after metadata as views

Bborie Park dustymugs at gmail.com
Fri Nov 30 08:44:52 PST 2012


On Fri, Nov 30, 2012 at 8:43 AM, Mateusz Loskot <mateusz at loskot.net> wrote:
> On 30 November 2012 16:24, Bborie Park <dustymugs at gmail.com> wrote:
>> On Fri, Nov 30, 2012 at 8:16 AM, Sandro Santilli <strk at keybit.net> wrote:
>>>
>>> You mean the change in the serialized value ? Really ?
>>> 16*2 bits enforced on all values for a "corner" case ?
>>>
>>
>> You're correct to question whether or not this is even worth making a
>> serialized format change.  I'd much rather see a format change for
>> ticket #2111.
>>
>> You can minimize (or in some cases eliminate) the amount of padding
>> that is done by wisely using tile sizes that leave little or no
>> remainder (raster width / tile width or raster height / tile height).
>
> Let's not forget the choice of arbitrary block (tile) size is also matter of
> optimisation for performance.
> It is a good idea to not to constraint that choice.
>

True.  The balance between cleanly divisible tile size and "smaller
tiles are faster" is hard to achieve.



More information about the postgis-devel mailing list