Layer Virtualization and SDE

[Howard Butler]

Jani and I chatted over IRC and decided that we would just put this 
function at the top of mapsde.c (and not include any prototype in 
map.h)  Not every layer provider needs to do something like this, and 
Jani said that it is easier to add stuff to the layer virtualization 
than to remove unnecessary things.

Howard

At 12:26 PM 12/8/2005, Jani Averbach wrote:

> > -----Original Message-----
> > From Howard Butler
> > Sent: Thursday, December 08, 2005 11:06 AM
> > Subject: Layer Virtualization and SDE
> >
> > Attempting to compile up mapsde.c with USE_SDE defined in a post-RFC
> > 3 world has created an issue.  SDE has a function in it to return the
> > index column for the table that someone has requested.  This function
> > is called msSDELayerGetRowIDColumn.  This function prototype used to
> > exist in map.h with all of the rest of the SDE functions.
> >
> > After the virtualization, this prototype was removed from
> > map.h.
>
>Sorry about that.
>
> > This causes things not to compile, but the larger issue is
> > that should there be a place holder for this type of thing in the
> > layer virtualization?  Would other data providers potentially have a
> > need to return which column is the index?
> >
> > Please give me some advice on what to do.  Short-term, I can just
> > stick the prototype in map.h and be on my way, but I'd rather things
> > behave uniformly.
>
>It looks like that consensus is heading toward to turn SDE off.
>We could extend the vtable if and when SDE is revved with the all 
>other stuff,
>so also my +0 would be for turn it off. (Adding new stuff to vtable is easier
>than removing already existing interface).
>
>BR, Jani