[Mapserver-dev] Re: OGR and SHAPEPATH ?

[Daniel Morissette]

After sending the message below this morning I found bug 295 in which we 
already discussed this and in which Steve explained that for shapefiles 
MapServer already checks relative to SHAPEPATH first, and if not found 
then tries relative to the .map file location.

I have made the same change for OGR (check under shapepath first and if 
not found try relative to .map file path) in the 4.1 development version 
(CVS) and updated the OGR-HOWTO.

Daniel



Daniel Morissette wrote:
> Frank Warmerdam wrote:
> 
>> Lars V. Nielsen wrote:
>>
>>> Hi Frank,
>>>  
>>> About half a year ago we touched on the subject of specifying 
>>> location of OGR / MapInfo TAB files.
>>>  
>>> I remember opting for making it possible to specify location relative 
>>> to the SHAPEPATH entry rather than to the MAP file, which would make 
>>> more sense to me (shape files and TAB files are equivalent entities). 
>>> And I remember you (and Daniel??) saying that you would look into 
>>> making it possible.
>>>  
>>> I've just checked the issue in the 4.0 documentation, and in 
>>> http://mapserver.gis.umn.edu/doc40/ogr-howto.html I found that "... 
>>> The SHAPEPATH is ignored for OGR datasources. ...". Does this mean 
>>> that it was never implemented as described above, or hasn't the doc 
>>> been upgraded ?
>>
>>
>>
>> Lars,
>>
>> There was alot of work done on path handling in mapserver but I am not 
>> exactly
>> sure of the current state.  I am cc:ing Daniel and Julien who are 
>> likely to
>> know more about this.
>>
>> Best regards,
>>
> 
> No this was never changed.  OGR data sources are still relative to the 
> mapfile path (or absolute).  I am still hesitant to change that since 
> apps that assume that the path is relative to the mapfile would be 
> broken if we evaluate the path relative to shapepath.
> 
> We could possibly try both, i.e., try relative to the mapfile path 
> first, and if not found then try relative to shapepath, but I don't like 
> this kind of magic because this often leads to confusion.
> 
> I'll CC the mapserver-dev list... if everyone agrees that OGR data 
> source paths should be evaluated relative to the shapepath *instead* of 
> relative to the mapfile path then I can make that change in 4.1.
>