[mapserver-dev] Re: [mapserver-users] Using e00 files

[Howard Butler]

On Feb 2, 2010, at 12:20 PM, Paul Ramsey wrote:

> Yeah, would be nice.
> 
> Howard did an RFC some time back.
> http://svn.osgeo.org/mapserver/trunk/docs/en/development/rfc/ms-rfc-37.txt
> 
> I think it could be whittled down to
> 
> if (! layer->projection)
> msLoadProjectionFromSource(lyr);
> 
> The odd/unfortunate thing about that is setting the projection becomes
> a kind of side effect of msDrawMap, probably. At least, the current
> workflow is such that the data sources aren't opened until "things
> start happening" (queries, map drawing, whatever). So either we
> piggyback the source projection setting into there, or we stick it
> onto the end of msLoadMap. But putting the projection load at the end
> of msLoadMap could make for some interesting problems for mapscript
> people who are loading a map file with invalid data entries and then
> making them valid before invoking an action.
> 
> The question of "where in the workflow to do the source projection
> reading" seems like the key one for this problem.

The proposal sought to enhance the layer providers ability to optionally return the projection to give us some flexibility in that regard.  Some types might not even need to have the layer opened to return the projection.

As 6.0 approaches, maybe this proposal can have more legs.  RFC 37 currently bends over backwards to ensure that it can preserve past behavior without changing mapfiles (no reprojection unless manually specified).  With 6.0, we could flip that to opt-out instead of opt-in and simplify the RFC quite a bit.  This work would be a nice sprint-sized chunk for a couple of people to chew on in a day or two as well...

Howard