time series problem 3 (Bug!?)

[Jeff de La Beaujardiere]

Yewondwossen Assefa wrote:
>  From the specs (wms 1.1 annex c c.3), the extent could be defined in 4 
> ways :
[most of which are not directly supported in Mapserver.

If you write a CGI script that does a bit of preprocessing and then
calls Mapserver, you can support all forms of the type extent (and
support additional dimensions if desired).  Parse the TIME argument,
then include parameters in the QUERY_STRING that correspond to replaceable
values (%value%) in your mapfile.  Your mapfile will need
to specify an appropriate value for METADATA wms_timeextent so that
the GetCapabilities response will include the proper extent, but
as far as I can tell that value is just a string that is sent as-is
by GetCapabilities and does not suffer from the same limitations
as the TIME value in the GetMap request parameter.

Regarding data every 01, 11 and 21 day of the month: because many
WMS clients have poor support for time[*], you might have best success
enumerating each value explicitly in a comma separated list.
   2005-01-01,2005-01-11,2005-01-21,...,2005-10-11
It would also be legal to say
   2005-01-01/2005-10-01/P1M,2005-01-11/2005-10-11/P1M,2005-01-21/2005-09-21/P1M
but many clients might choke on that list of interleaved ranges.

-Jeff DLB

[*] We should have made it more clear in the specification that
"optional for dataset" does not equal "optional for client." -Ed.