[gdal-dev] Re: OGR FileGDB driver: Failed at creating table ... (General function failure.)

[Hermann Peifer]

On 16/09/2011 16:49, Paul Ramsey wrote:
> On Fri, Sep 16, 2011 at 3:16 AM, Hermann Peifer<peifer at gmx.eu>  wrote:
>> Frank, Even, Paul, et al.:
>> Could it perhaps make sense if the morphToESRI() function paid more
>> attention to these string identifiers in order to produce even more
>> "ESRI-friendly" WKT?
>
> Yes, probably we will have to. I wonder how we would go about figuring
> out the relevant differences...? In the meanwhile, since you've found
> that using the WKID works perhaps in the FGDB driver we can try and
> avoid using the WKT at all when we have a WKID available.
>
> P


Isn't the WKID a numeric identifier, and not the string identifier I was 
mentioning in my previous mail (?)

My little exercise only showed that with an appropriate string in the 
WKT (here: "ETRS_1989_LAEA"), the "General function failure" can be 
avoided when creating out.gdb through the FileGDB driver. It doesn't 
mean that the string value "ETRS_1989_LAEA" is enough to define a valid 
spatial reference.

When opening the generated out.gdb in ArcCatalog, I get a "General 
function failure", "Cannot read SRS..." failure. AFAICT: in order to 
create a meaningful spatial reference, the complete WKT has to be passed 
over: morphed string identifiers, morphed projection and parameter 
names. etc.

Hermann