RE: Preview of full latest Java Interface : Comments on SimpleService interface

+1 to removing the word Simple and streamlining the overloaded methods, for the same reasons given...
Kevin

-----Original Message-----
From: public-ddwg-request@w3.org [mailto:public-ddwg-request@w3.org] On Behalf Of Jo Rabin
Sent: 27 February 2008 09:00
To: public-ddwg@w3.org
Subject: RE: Preview of full latest Java Interface : Comments on SimpleService interface



I share your concerns on most of this. In particular I think the method names are unmanageable (and actually self contradictory - did I mention that this is the "Simple" interface, maybe better stick the word Simple in again just in case). I think we could banish the word Simple from the interfaces as Jose suggests and leave it in the package name as a clue.

We have a frighteningly large number of overloaded methods. One for every day of the week, two for Sunday, and a spare for leap years.

We have also a SimpleService.newSimplePropertyRef(String aspect, SimplePropertyName) method that appears to duplicate the SimplePropertyName.newSimplePropertyRef(String aspect) method. So I think we should take the factories out of SimplePropertyRef and SimplePropertyName interfaces as being redundant.

All that aside, my biggest concern is that we are in a process of taking two steps forward and one step back. I can see the pages of the calendar blowing away, like in those old movies, and I am planning to be on a plane, Saturday, to attend a meeting that will resolve this document as at least a FPWD and preferably a LCWD. So in short, this process needs to come to a conclusion.

Jo

> -----Original Message-----
> From: jmcf@tid.es [mailto:jmcf@tid.es]
> Sent: 27 February 2008 08:27
> To: Jo Rabin
> Cc: public-ddwg@w3.org
> Subject: Re: Preview of full latest Java Interface : Comments on 
> SimpleService interface
> 
> + I'm missing a newHttpEvidence() method. We agree that the Java 
> + binding
> is gonna also have a newHttpEvidence(Map) but the method with no 
> parameters is also needed as it is the only method present in IDL and 
> in every language binding
> 
> + The methods that only receive a property and not an aspect, Are they
> going to return the property value for the default aspect of the property?
> 
> + The getPropertyValues method should have at least one more 
> + overloaded
> method:
> 
>     public SimplePropertyValues getSimplePropertyValues(SimpleEvidence
> evidence,
>             PropertyName property) throws NameException, 
> SystemException;
> 
> which will return all the property values for that property in all the 
> aspects of that property
> 
> + I'm a bit worried about the complexity of the overloading that we 
> + are
> putting in place We should think of what is gonna be the most common 
> use case and model it in a convenience function and leave the other 
> cases behind the SimplePropertyRef and SimpleProperty interfaces
> 
> So perhaps it only makes sense to mantain the 
> getPropertyValue(Evidence, String localPropertyName) as the only fast 
> food method ...with the possible addition of the methods that allow to 
> retrieve all the properties of an aspect
> 
> Also I'm not very convinced that we need to have in everyplace the 
> word 'Simple', that could be left in the package name ...
> 
> Best Regards
> 
> Jo Rabin escribió:
> >
> > I've uploaded an edited copy of the latest Java interfaces as I 
> > believe agreed on list. However, there are a couple of instances of 
> > editorial discretion wrt names etc.
> >
> >
> >
> > Seeing as I am very short of time I'd like people to review this and 
> > make any comments known _/before/_ I go ahead and edit/issue a new 
> > draft. I'm going to hold off producing the draft till tomorrow.
> >
> >
> >
> > You will find the new interfaces at
> >
> >
> >
> >
> http://www.w3.org/2005/MWI/DDWG/drafts/api/simple/java/org/w3c/ddr/sim
> ple/
> >
> >
> >
> > Jo
> >
> >
> >
> >
> >

Received on Wednesday, 27 February 2008 09:20:05 UTC