W3C home > Mailing lists > Public > www-qa-wg@w3.org > January 2005

First stab at LC issues analysis

From: Dominique HazaŽl-Massieux <dom@w3.org>
Date: Wed, 26 Jan 2005 10:38:59 +0100
To: www-qa-wg@w3.org
Message-Id: <1106732339.5692.157.camel@stratustier>

Here come a few thoughts about our LC issues list [1]:

#1052 "Classes of product" unclear and dangerous
LH and DM commented on this, thinking it was maybe now unclear, but
certainly not dangerous; I think a good way to understand the XML Core
WG concerns would be to come up with a list of classes of products for
xml:id and see with them if this would indeed be too restrictive.
In any case, the confusion shows that we need to better explain what we
mean by classes of products, maybe with examples less "obvious" as the
ones currently in SpecGL (although Ruby looks very much similar to the
xml:id case).

#1041 Conformance is not a yes/no proposition (wrt filling an ICS)
IanH pointed that an Implementation Conformance Statement would imply
that Conformance was a yes/no proposition, whereas it can only be
partially proved by the intermediary of tests; my take on this is that
there is a misunderstanding as to what the ICS is about, which means
again that we need to clarify it - an ICS is not saying that you
implement in a conformant-way this or this feature, but that among the
various possible flavors defined in the spec, you've implemented this
one or that one. Is that a fair explanation?

#1045 Avoiding device-dependent profiles
I think the gist of the issue relates to our general warning wrt
variability; reading the thread in www-qa actually showed that
apparently IanH thought we were actively suggesting profiling
specifications, which I think goes against what most of us think... 

#1047 Addition to error mechanism
The suggestion (that defining the spec so that no (or almost) no error
is possible) seems good to me; let's see if nobody disagrees with it,
and if so, make it an editorial matter.

#1048 Additions to "write tests"
"another technique is to go back and create new tests for old sections
once the sections are better understood and more mature."
"one must absolutely check the _interactions_ of assertions"
Both seem reasonable to me.

#1049 Formal vs prose language normativity
We probably need to address the case where both prose and formal
language have the same type of expressiveness; I'm not sure in what way,

#1050 Modesty requirement
"Specifications should not claim to be simple, easy, device-independent,
conformant to WAAA or QAG, or make other claims about their quality or
conformance to other specifications"
I think I disagree with this; more to the point, I think we do want
specs to claim conformance e.g. to SpecGL.

(I've left the other issues out of this analysis, because I thought
there were either editorial or their status was not clear to me yet)



Dominique HazaŽl-Massieux - http://www.w3.org/People/Dom/

Received on Wednesday, 26 January 2005 09:39:01 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 22:43:38 UTC