[postgis-devel] "userland" schema in raster regression tests
dustymugs at gmail.com
Mon Dec 19 07:35:06 PST 2011
On Mon, Dec 19, 2011 at 7:29 AM, Sandro Santilli <strk at keybit.net> wrote:
> On Mon, Dec 19, 2011 at 06:57:45AM -0800, Bborie Park wrote:
>> "make check" does work for me on all my platforms. The problem arose
>> because I'm checking the values present in the raster_columns and
>> raster_overviews views without defining the columns to check.
>> One of the columns being grabbed was the schema where the test raster table
>> resides, which on my setup defaults to "userland" instead of "public".
> "make check" should create its own setup, that's what I'm saying.
> There should be no difference between "yours" and "my" setup when it
> comes to running "make check". If there is, it is a bug in the system
> creating the isolation.
> Isn't it the .sql file under regress/ which is creating the test raster
> tables ?
Yes, the sql creates the test raster tables.
In my setup though, no template database has the schema "public" and
the search_path itself has no "public" in it. So, when postgis_reg is
created, it won't have a "public" schema. Rather it'll use what is in
template1, which is "userland".
And since raster_columns is the catalog of rasters in the postgis_reg
database, my initial use of "SELECT *" grabbed whatever schema the
test raster tables were located in.
More information about the postgis-devel