[standards-jig] Disco features caching

Jacek Konieczny jajcus at bnet.pl
Wed May 21 16:51:31 UTC 2003


On Wed, May 21, 2003 at 09:09:26AM -0600, Peter Millard wrote:
> Jacek Konieczny wrote:
> [Stuff Munched]
> > That is why some kind of disco reply caching should be implemented on
> > server. It could be just caching of last client reply - but this will
> > not work well if the feature list changes during a session. Other
> > solution would be publishing the list to the server (similary to what
> > can be done with disco items). The list on the server should be
> > maintained for one resource and one session - I don't think persistent
> > storage of this information is needed.
> 
> This seems like an implementation issue for servers which are designed to
> interact with mobile clients. How would this impact protocol?? (I don't think it
> should).

Specs should say if the disco reply may be cached, if it may be assumed
to be always up-to-date, etc. 
It will be no good special servers with proprietary extension would be
required for good work with mobile client.
When specs would say nothing about that and server would programmer
assume, that it may be cached and implement caching, client programmer
could assume that it never is cached and. eg. reply with different
feature list to different users, then that client and that server whould
not work well together. 

Greets,
        Jacek 



More information about the Standards mailing list