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

Re: TOC checkpoints

From: Alex Rousskov <rousskov@measurement-factory.com>
Date: Fri, 25 Oct 2002 22:27:23 -0600 (MDT)
To: Lynne Rosenthal <lynne.rosenthal@nist.gov>
cc: www-qa@w3.org
Message-ID: <Pine.BSF.4.44.0210252144200.77136-100000@measurement-factory.com>

On Fri, 25 Oct 2002, Lynne Rosenthal wrote:

> After some thought - I believe that the "Include a TOC" checkpoints are
> really techniques and not a checkpoint.


> The goal is:
> LH>Our goal is this:  it should be easy for the user to find all of the
> LH>information necessary to understand the conformance policy and all of
> LH>its details, regardless of spec partitioning.
> and the checkpoint should reflect this.

I agree. It is a difficult checkpoint to write. On one hand, the
checkpoint cannot say "conformance policy subjects/elements should be
EASY to find" (general but not be testable or practical). On the other
hand, the checkpoint cannot require a specific technique like TOC
entries (testable and practical but not general).

You may be able to resolve the conflict by replacing informal "easy"
with a more formal (albeit less general) characterstic, like "time".
For example,

	Location of any single spec object MUST require
	neglible absolute time or neglible time relative
	to location of all spec objects.

In other [informal] words, "easy" means "much faster than looking
through most of the spec".

The above is far from perfect, of course. Negligble can be defined if
you want to get even more formal (you will need to definitions, one
for absolute as in "at most 15sec to find a term in a 1-page spec",
and one relative as in "at least 100 times faster to find a link in a
10-page TOC than to read 1000-page spec"). But I doubt it is necessary
to be that formal.

"Spec object" should be defined as anything the spec defines, any
"other spec" the spec is referencing, and any "other spec object" the
spec is using without a definition. Did I miss anything?

> So, I would like to see 1 checkpoint that captures the goal and have
> that checkpoint in 1 place. This eliminates the need to repeat the
> checkpoint in every DoV-related section, it ensures that any non-DoV
> items that could affect the conformance policy would also be covered
> by the checkpoint, and we don't need to worry about specs that may
> not have a TOC.

There is no need to limit this checkpoint to conformance policy, IMO.



                            | HTTP performance - Web Polygraph benchmark
www.measurement-factory.com | HTTP compliance+ - Co-Advisor test suite
                            | all of the above - PolyBox appliance
Received on Saturday, 26 October 2002 00:27:27 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 20:40:30 UTC