W3C home > Mailing lists > Public > www-qa@w3.org > October 2002

Re: Defining "public interfaces" in specifications

From: Alex Rousskov <rousskov@measurement-factory.com>
Date: Wed, 30 Oct 2002 09:36:34 -0700 (MST)
To: Eric van der Vlist <vdv@dyomedea.com>
cc: www-qa@w3.org
Message-ID: <Pine.BSF.4.44.0210300925280.29121-100000@measurement-factory.com>

On 30 Oct 2002, Eric van der Vlist wrote:

> Yes, that's what I meant saying that unique identifiers would be
> given to these public interfaces. It would become much easier to
> evaluate the impact of changing one them.

Any normative piece of a spec is a public interface. Unique (within a
spec) identifiers should be given to sufficiently large or important
normative pieces of a spec.  It should be up to the authors to decide
what is "sufficiently large or important". I hope SpecGL already has a
general requirement for labeling normative pieces. If SpecGL does not,
it must be added.

It would be desirable for identical concepts in two specs (e.g., two
versions of a spec) to have identical identifiers, but we should not
require that. I think we already agreed that normative pieces should
be "easy to find" in a spec. That's sufficient. It is impossible to
automate backward compatibility verification process anyway.

I see no need to introduce a new concept of "public interface" or a
new "public interface registry". Your examples do not show [me] how
these new things will help or even how they would make any difference.

Alex.

-- 
                            | HTTP performance - Web Polygraph benchmark
www.measurement-factory.com | HTTP compliance+ - Co-Advisor test suite
                            | all of the above - PolyBox appliance
Received on Wednesday, 30 October 2002 11:36:40 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Sunday, 6 December 2009 12:13:59 GMT