Re: SpecGL editorial comments

At 01:00 PM 12/19/03 +0100, Dominique Hazaël-Massieux wrote:
>[...]
> > CP2.1:  "If your class of product matches one or more terms in the
> > list..."  Can any given CoP match more than one term, as this implies?  (I
> > dunno' ... this is just a question that occurred to me when I read this.)
>
>Well, I guess it would be easy to create requirements for a
>consumer/producer type of products.

Where I'm getting confused is ... "your CoP matches more than one term" 
versus "a product represents more than one CoP.

A *product* could be a producer and it could be a consumer (e.g., it could 
be capable of generating SVG, and it could be capable of reading/viewing 
SVG).  This product should be subject to the conformance requirements for a 
viewer (CoP), and it should be subject to the conformance requirements for 
a generator (another CoP).  This product does not necessarily imply a 
third, producer+consumer CoP.

Note that I'm not talking yet about an SVG Editor.  An SVG Editor is a 
viewer, plus a generator, plus a ... well, an editor that alters the 
consumed content according to some rules or interface, before producing 
altered content.  That third component ("altering") is important, and 
potentially the subject of conformance requirements.

So I would think "SVG Editor" (or more generally, "content editor") could 
be a CoP.  (Btw, I don't see a CoP in the list that matches "editor").  But 
does this editor CoP match more than one term in the list?  Or is it really 
a single CoP that either matches a single CoP in the list, or needs to have 
its (single) CoP defined?

Am I making any sense, or is this just some pre-holiday gibberish?


> > CP3.1, Examples:  I cannot figure out what is meant by "...only one 
> profile
> > can be implemented at a time."
>
>Well, you could imagine that it would be forbidden to implement both a
>tiny and a mobile profile in an SVG implementation (why would you do
>that, I don't know, but maybe that helps understanding the meaning :)

I still don't understand.  Let's look at two types of implementation, a SVG 
generator and a SVG viewer.

1.) SVG generator implementation could have a user-selectable mode to 
either emit Tiny-conforming content or Basic-conforming content.

2.) SVG viewer implementation could be capable of viewing valid Tiny 
content and viewing valid Basic content.

I cannot understand why one would want to impose a constraint against #1, 
or a constraint against #2.

Maybe we need a different "for example" here -- something common and useful 
whose purpose and utility are clear -- unless I'm missing something really 
obvious.


> > CP3.3, ConfReqs:  The not-applicable exemption is missing (if profiles are
> > not used).
>
>Hmm... Since the ConfReq starts with "If derived profiles are used", I
>don't think it's useful to have more than that for exemption.

Ah, we talked about this in telecon.  I wasn't sure that "if derived 
profiles are used" was redundant with the (missing) exemption 
sentence.  The missing exemption sentence would be "This CP is n/a if 
profiles are not used."  That would basically be equivalent to and 
redundant with prefacing the ConfReq sentence with "If profiles are used...".

I.e., If we started the ConfReq with, "If profiles are used and derived 
profiles are allowed...", that would obviously make the missing exemption 
sentence redundant.


> > GL5:  "implementation dependent values (or features)" -- definition is
> > garbled.
>
>Can you clarify what you mean by that?

Now that I reread it, I understand it.  (Commas surrounding "or the 
behavior of a product that implements a feature" might help.)


> > CP5.3, statement & ConfReqs:  There are 3 occurrences of "items".  Isn't
> > this CP specifically about "discretionary choices", which is one of the
> > three types of "discretionary items"?   If "yes", then the 3 
> occurrences of
> > "items" are incorrect -- change to "choices".  (If "no", then this becomes
> > a substantive issue, not editorial -- this CP used to be about
> > discretionary choices and seems to have mutated).
>
>It was a decision taken by the WG, I believe, when we adopted
>DavidM/Lynne's proposal of rewrite of this CP.

Hmmm... then I really don't understand what's intended with some of these 
CPs.  (That in itself might suggest that more clarification is needed.)

We have a lot of documented issue history that suggests "items" is 
incorrect.  Can someone please point me to subsequent issue history that 
documents the change?

(I think there may be substantive issue here.)


> > CP9.2, Discussion:  it doesn't relate very well to the subject of the 
> CP or
> > to the ConfReqs.  If there *is* a connection, there should be a little 
> more
> > verbiage to expose the connection.
>
>Agreed; could you propose some verbiage to make this connection?

Okay (might be good to queue an AI for me, so that I don't lose it.)


> > CP9.3, Discussion:  what does "relevant assuring parties" mean?
>
>e.g. a certification authority?

I don't know.  If that's what it means, I guess it should be defined.


> > all CPs:  each CP in CR version of SpecGL points to a SpecET section that
> > is paired with (points back to) the preceding *WD* SpecGL version.
>
>I don't think there is much we can do about that now.

SpecET is in /QA/WG/ space and can be fixed.  (I just had to do this for 
OpsET, because of the 2nd publication of the OpsGL CR text a couple weeks 
after the first publication.)


> > Sec 4.5:  notice the nasty formatting glitch of the blockquote in the
> > Conformance Claim example (unless your browser window happens to be just
> > the right width).
>
>I don't see any formatting glitch; could you give more details?
>(browser, possibly a screenshot, etc.)

It happens in any browser.  It's badness depends on the browser window 
width.  It is because the stuff after "Example" is right-justified instead 
of ragged-right (and the long URL cannot be broken).  Attached is a 
particularly bad view (from Amaya).

-Lofton.