W3C home > Mailing lists > Public > www-qa@w3.org > June 2003

TA spin-off issues [was Re: Draft Minutes QAWG Telcon Monday 09 June 2003]

From: Lofton Henderson <lofton@rockynet.com>
Date: Wed, 11 Jun 2003 16:46:24 -0600
Message-Id: <5.1.0.14.2.20030611162940.03411ec0@rockynet.com>
To: <www-qa@w3.org>


My ideas about the three dangling test assertion details:

At 02:39 PM 6/11/03 +0100, Andrew Thackrah wrote:
>[...]
>         Three spin-off issues are identified that should be addressed in 
> future meetings:
>
>         1 Priority
>         2 Creator (spec author | test developer )
>         3 Auto-generation (spec->test via formal grammar)

Recall that we are talking about test assertions (TA) in the SpecGL 
context, i.e., we resolved that we think there is value to specifications 
if some set of test assertions are developed before publication of 
Rec.  This is independent of TA considerations that might (will) arise in 
the context of TestGL.

Clear and specific "conformance requirements" are of principal importance 
in the SpecGL context (and we think they are reflected in GL13).  But there 
is also value to having TA developed concurrently.

Priority
-----
CP14.1 and 14.2 are now P2.  Our priority definitions 
are:  P1--critical/essential; P2--important/desirable; 
P3--useful/beneficial.  I think P2 is right for the SpecGL context, but I 
don't feel real strongly.  P3 would work for me also.

Creator
-----
There was a lot of sentiment that there is benefit from having 
participation by people *other than* the spec authors.  At least a 
collaborative effort.  I would say that "others" is desirable, "authors" is 
acceptable.  I would also relegate this detail to SpecET.  SpecGL now says 
"provide", and that leaves the who&how wide open.  So I suggest that SpecET 
discuss it, recommend "other" or "author-other" collaboration, with 
"author" being acceptable (better than nothing).

Auto-generation
-----
I don't feel real strongly.  But I think, in the SpecGL context, that 
nothing is lost by exempting auto-generation scenarios.  In the TestGL 
context, we could decide (or not) that each test case must have an 
associated test assertion, even if one has to be constructed artificially 
in between the formal grammar of the spec and the auto-generated test cases.

Regards,
-Lofton.
Received on Wednesday, 11 June 2003 18:45:41 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Sunday, 6 December 2009 12:14:00 GMT