[MOBY-l] service = method ?

[David Block]

On Monday, October 28, 2002, at 11:47  AM, Martin Senger wrote:

> It seems that my question attracted some attention but to be fair I do 
> not
> understand some answers we have received. Very probably the fault is 
> mine
> - I did not express myself clearly. Please allow me to ask it again and
> simply:
>
> Can a Moby central register a web service that has more tha one method?
> Does it want to do it? And can such methods be dependent on each other 
> (in
> a sense that one must always be called before the second one)?
>
In other words, does Moby contain an option of saving session state?  If 
so, what mechanism does it use, and what does that mean for client 
programming?

Both cgi.pm and java servlets allow state to be maintained fairly 
transparently, so we should not have to look too far for 
implementations, and we should definitely not roll our own here, as we 
want others who have more of a vested interest in session security to do 
the development here for us.

I recommend following closely what IBM and MS are doing with Web 
Services Security, which should become a de facto standard for this kind 
of thing.

Martin - I understand your point.  Having each client roll their own 
session ID variable/cookie/whatever solution is clearly a worst-case, 
unmaintainable solution, and it raises the bar for new services.  We 
should have some way of "use Cookie" or "import javax.servlet.*" or 
whatever and just roll with someone else's solution.  The most important 
thing then, is to come up with an interface, and then a reference 
implementation that others can copy.

And here's where I'll back out again, as I'm too rusty on Moby internals.

Watching with interest,
Dave

> The example I gave about timely jobs was perhaps misleading. Because the
> situation does not need to deal with time-consuming jobs, but with the
> whole "bussiness logic". For example, a bibliographic query service may
> want to refine queries step-ny-step, each time getting more precise
> results of the query. Such example would easily require to have one 
> method
> for starting the query and yet another method for continuation queries.
>
> And by the way, the Moby central itself is an example of a web service
> that has always two dependent methods (new() and a real one). Therefore,
> if Moby central accept only one-method services, we will not be able to
> register the Moby central itself in Moby central (I am not saying that I
> ask for it but I am just playing with the idea...).
>
>    Martin
>
> --
> Martin Senger
>
> EMBL Outstation - Hinxton                Senger at EBI.ac.uk    
> European Bioinformatics Institute        Phone: (+44) 1223 494636     
> Wellcome Trust Genome Campus             (Switchboard:     494444)
> Hinxton                                  Fax  : (+44) 1223 494468
> Cambridge CB10 1SD
> United Kingdom                           
> http://industry.ebi.ac.uk/~senger
>
>
David Block                                  dblock at gnf.org
GNF - San Diego, CA             http://www.gnf.org
Genome Informatics / Enterprise Programming
Weblog:      http://radio.weblogs.com/0104507/
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: text/enriched
Size: 3309 bytes
Desc: not available
URL: <http://lists.open-bio.org/pipermail/moby-l/attachments/20021028/39233cf1/attachment-0001.bin>