[mapserver-dev] IGNORE_MISSING_DATA

[Stephen Woodbridge]

Paul Ramsey wrote:
> All,
> 
> Is there a reason that IGNORE_MISSING_DATA is a compile-time option? I
> have a client request to make it a run-time option, and also add the
> ability to run-time configure to error out in the case of WMS client
> layers that fail.
> 
> I can RFC this, though it's relatively small... IGNORE_MISSING_DATA
> would go away as a compile time option, the run-time behavior would
> remain as the default (fail on missing) and there'd be a couple of
> PROCESSING directives to fail (or not) on missing data and failed WMS
> client layers.
> 
> Thinking aloud, there's a number of ways data can be missing, in a
> number of places
> 
> -- wms client layers
> -- wfs client layers
> -- missing data files
> -- missing files referenced by tile indexes
> -- bad database connections
> -- failed SQL statements
> 
> Establishing and adding global policy would involve touching a great
> deal of code. Hm. Unfortunately, I only have interest in the
> tile-indexed-rasters and wms clients areas.
> 
> Anyone else have interests in this field of endeavor?

Paul,

I use this feature extensively for tile-indexed-vectors. Basically, my 
tileindex points to county directories, but in any given directory you 
may or may not have any specific file. So the mapfile looks like:

LAYER
   NAME "minroads"
   TILEINDEX "county-index"
   DATA "minroads"
   ...
END

LAYER
   NAME "majroads"
   TILEINDEX "county-index"
   DATA "majroads"
   ...
END

etc for each POSSIBLE data layer that might be available in a county. If 
the data layer does not exist, then it should be ignored.

-Steve