[gdal-dev] Change to SetMercator and SetMercator2SP creates WKT without Latitude of Origin parameter

[Chris Eustace]

Frank Warmerdam wrote:
> Chris Eustace wrote:
>> This question is probably directed to Frank.
>>
>> Our autotests here at work have shown there are differences between 
>> the CRS WKT strings generated from GDAL v1.4.4 to GDAL v1.5.1.
>>
>> I've scanned around a bit and noticed that ogrspatialreference.cpp 
>> has altered between these versions.
>>
>> I did a bit more hunting and found that the GDAL changeset number 
>> concerned is 12207. The subject relates to moving sfc/oledb stuff for 
>> the move to spike.
>>
>> My question is, whether the changes to the SetMercator calls are 
>> intended changes to ogrspatialreference.cpp or not? What we are 
>> seeing here is that if "dfCenterLat" is 0.0, that Latitude Of Origin 
>> parameter is no longer present in the resultant WKT.
>
> Chris,
>
> Hmm, that is funky.  It clearly has nothing to do with the migration of
> oledb stuff to spike.  It must have got caught up in a big commit by
> accident.  I'm not sure what the rationale for it was.
>
> Is this ommission causing problems for something?  Generally GDAL itself
> will assume zero values for parameters like this if they are omitted.
>
> I assume at the time I was looking into some issue with software packages
> not expecting this parameter for Mercator (perhaps ArcGIS?) and was
> experimenting with only including it if it was non-default, but I don't
> recall in any detail.
>
> I would suggest you file a ticket on the issue, and include any notes
> on why it is causing you problems.  I can at least check ArcGIS PE WKT
> and see if it includes this parameter.
>
> Best regards,
We had a close look here as to where and how we use the CRS WKT. For us, 
everything comes in and goes out through GDAL so, as you correctly say, 
there is no problem as any missing parameters will have zero assumed for 
their values.

So, no issue for us and I'm therefore not planning to file a ticket. It 
was just that our unit autotests picked up a difference so I was more 
after a clarification - especially since your "funky" changeset did not 
seem to represent that particular change.

So, all good here.

Thanks for your time,
Chris Eustace.