RE: [RTF] AC019 proposal to WSA WG

><sd> I agree with you on "stable" and "evolvable."
>"Reliability," in the mechanical world will include time factor
>because of ageing, though in software, I am inclined to think
>time is not a factor. We added "over time" for clarity, though
>it may be redundant from some perspectives.
></sd>

OK. M-W defines reliability as "giving the same result on successive 
trials". To me, this implies time dependence. But this is a nit
and if 'over time' is clarifying for some people, that is OK with
me.

><sd>
>I guess the wording is not descriptive enough.
>The idea is that the availability of a web service ("dinner available
>only between 7PM-8PM") will be "conveyed" to service 
>requesters whenever
>possible so that they do not hang ("asking for dinner at 
>4AM"). We did not
>think we should REQUIRE all web services to announce their 
>availability, and
>
>hence "where possible." We will think of an alternate text.
></sd>

I understand better now. How about:
Web Service Architecture will enable a Web Service to describe 
its availability constraints.

>
>Also, reliable messaging is useful for all 3 of the interactions with 
>a Web Service you have described. Is there any requirement for
>specifically applying reliability to invocation/execution?
><sd> 19.1 includes discovery, access and invocation, and 19.3 
>applies to all
></sd>

OK - I just thought you might have a specific requirement for reliability 
of execution along the lines of "enables a Web Service to produce the 
expected output on successive trials through the support of a well defined 
versioning mechanism"

>
>>
>>AR019.2 The Web Services Architecture enables a conforming Web Service
>>implementation to be stable with respect to its definition.
>>	D-AR019.2.1 A Web Service can be defined independent of its
>>implementation.
>>
>
>I don't see how this achieves stability. This seems to me to belong 
>more in the proposed 'loose-coupling' CSF.
><sd>
>I am unaware of "loose-coupling" CSF - please provide pointer to the
>requirements
>document. If this is stated elsewhere, we will simply refer to it.
></sd>
>

This was to be discussed in the last telecon. See [1]. Hopefully it will 
get some air time next teleconf.

[1] http://lists.w3.org/Archives/Public/www-ws-arch/2002Jun/0238.html

Received on Tuesday, 9 July 2002 17:39:53 UTC