[OSM-dev] API 0.5 is on the way

Frederik Ramm frederik at remote.org
Tue Sep 18 10:38:56 BST 2007


> The problem is not the language here. There is currently no way the
> validating XML-parser can know what kind of message it is parsing.  
> Thus
> it can not check a against the grammar of the specific message.

That's true but then again the Wiki page should be aimed at the human  
reader. I'll try and create an extra page with the full and complete  
DTD on it.

> The traditional solution is to have one DTD per message. That would
> solve the problem. Not sure that we want to go that way.

I'd say that we are mixing different levels here. There is ONE way to  
encapsulate any numer of OSM obejcts in an XML message, and we need  
one DTD for that. Any message returning OSM objects will adhere to  
that DTD and can be checked for formal correctness. Whether the  
message really contains the objects you were looking for is not  
something that can be checked on the XML level; that's one level up.

> For definition in the wiki that approach can be used. I added in the
> following table the specific OSM root element:
> http://wiki.openstreetmap.org/index.php/ 
> OSM_Protocol_Version_0.5#Basic_Methods_for_Object_Access_and_Manipulat 
> ion

Frankly, I think the table has become much less readable and I liked  
the previous version that just said "per above DTD" much better... it  
is true that simply saying "per above DTD" will force the reader to  
combine in his head the root element and the elements for the  
specific objects but as I said before, this table is intended for  
human beings to read.


Frederik Ramm  ##  eMail frederik at remote.org  ##  N49°00.09' E008°23.33'

More information about the dev mailing list