[OSM-talk] Strange api behaviour

[David Sheldon]

On Thu, Mar 16, 2006 at 02:15:50PM +0100, Immanuel Scholz wrote:
> >> Re: 0.2 retunring 0.3 stuff
> >
> > Just to confirm: does this mean that the 0.2 API is basically no longer
> > available?
> 
> This means, there is either a problem with some caches after the url got
> rewritten or the rewrite rules in the apache config of the server is
> messed up.
> 
> What it does NOT mean is that Steve is trying to run a 0.3 Api on the url
> namespace of 0.2. RIGHT? STEVE? YOU WOULDN'T??? ;-)

It means that with mod_ruby the dao object (used to generate XML) can
only exist once in the system. In this case it is the 0.3 one. 

The solution is to call it dao03 or 0.3/dao or something like that.

In a way it is a problem with caching, but not one that is going to go
away. It may manifest differently if steve restarts apache and someone
hits the 0.2 API first (in that the 0.3 URL will be returning 0.2 XML).

This is a major problem, and I'm sure Steve is working on it. When he
has a solution we might be able to develop on the development server
too.

What ruby probably needs is something like FastCGI or Persistant Perl,
rather than mod_ruby. The alternative would be to run each API in its
own web server and proxy them through www.openstreetmap.org at the
correct URL locations. This could also solve the development server
issues.
 

David
-- 
it is a strange, strange meme. people are either immune, or very seriously
infected. could it be a tailored meme-war plague? - TWIC, o.c.ousfg