Re: [pmapper-users] Search definition via XML file for DB

[Armin B. <arm...@gm...>]

On 24/01/2012 07:29, Andreas Douvalis wrote:
> Sorry for my mistakes! Again my question:
>
> Why in Search definition via XML we have to write dsn string?
>
> I think it is more convenient to construct dsn string from connection string of
> layer in map file.
> In other case I have to maintain two config files
> (map file and search.xml). Am I right or I miss something?

Just to clarify this: for standard search definitions you do not need to 
define the dsn, just for the special case of type="suggest" and 
type="options".

One reason for this was probably just simpler handling of the dsn 
without the need to extract it from the connection in the map file. 
Another was that you could use any type of DB connection in combination 
with any type of layer. Maybe looking a bit exotic, but I found use 
cases for it.

There needs to be also a definition of the DB encoding. This might be 
possible to be extracted from some Postgres system tables but I don't 
think this is too much asking to set it in the XML definition.

A simplified definition could e.g. be something like
   <dsn encoding="UTF-8">@</dsn>


If you find this double definition annoying create a "new feature" 
ticket in the p.mapper TRAC site suggesting reading the dsn from the map 
file in case of Postgis layers. Then the chances are much better that 
this feature once gets added.

Armin