Re: ISSUE-101 (us-ascii-ref): Chairs Solicit Proposals

On 03.03.2010 17:05, Sam Ruby wrote:
> "Spec reference for US-ASCII"
>
> Per the decision policy, at this time the chairs would like to solicit
> volunteers to write Change Proposals.
>
> http://www.w3.org/html/wg/tracker/issues/101
> http://dev.w3.org/html5/decision-policy/decision-policy.html#escalation
>
> If no Change Proposals are written by April 5th, 2010 this issue will be
> closed without prejudice.
>
> Issue status link:
> http://dev.w3.org/html5/status/issue-status.html#ISSUE-101

Below a few comments on the objections recorded in the poll:

 From Boris:

"I object to this on the grounds that referencing a pay-for spec for 
ASCII will essentially make the reference useless. No one is going to 
bother paying for this when free copies are so readily available. So if 
we are referencing the pay-for copy, we are effectively not helping 
people out at all."

What's relevant is that we reference a document that actually is 
normative. If it is "pay-for", that's sad, but doesn't change the reality.

In practice, almost all specs in the IETF cite ANSI for US-ASCII, and I 
have never seen any complaint about that, because, guess what, nobody 
needs to look it up anyway.

"And since we seem to think it is important to help people out using a 
reference (no one has suggested removing the reference), we should 
actually attempt to do so. Pointing to a reference that nobody will use 
does not fulfill that goal."

Well, that can be addressed by having an informative reference to 
something that has that information. I don't believe it's needed, though.

"I also object to making any changes here in general as this whole issue 
is a giant waste of time. If this time waste keeps happening I suggest 
we amend whatever needs amending to prevent further incidents like it. 
Since this change proposal was the one that caused this whole time 
waste, I'm choosing to object here."

I agree that it's bad that we have a multi-stage escalation for an 
editorial issue. This should have been addressed earlier. Much earlier.

 From Henri:

"I object to changing the reference to anything that cannot be obtained 
as plain text, HTML or PDF free of charge by issuing an HTTP GET 
request. (For avoidance of doubt, I object to referencing a .pdf.zip, too.)"

I agree with that goal. It doesn't change the fact what the normative 
definition of ASCII is, though. Also note that the alternate change 
proposal (using ECMA-6) actually addresses that.

 From James:

"Having said that I do not believe that anyone is seriously going to try 
to follow the reference in the spec to learn what ASCII is. This entire 
issue seems to be a waste of the group's time and I am disappointed that 
the chairs have allowed it to come to a poll when it purely editorial, 
there are outstanding issues of substance, and we have missed several 
deadlines and revised estimates for Last Call."

Indeed, it's unlikely that anybody actually needs to look it up. So why 
is it then a problem to use a proper reference?

And yes, it's bad we're wasting time of this. A simple way to avoid this 
would be to listen to those who opened the issue.


Henri:

"I object to this change proposal, because it changes the ASCII 
reference to something that doesn't define ASCII in terms of Unicode 
code points."

That's a good point, but doesn't change the reality that US-ASCII is not 
defined that way.

It certainly would be nice if we had something that is free, readable, 
easy to reference, and which defines US-ASCII (the character set) as the 
first 127 Unicode code points, and US-ASCII (the encoding) as the subset 
of UTF-8 mapping these code points to octets. I would support this, both 
as a separate spec (be it W3C or IETF) or inside HTML. But beware, there 
may be devils in the details with respect to certain control characters 
(which may not be relevant for HTML, but potentially elsewhere).

Best regards, Julian

Received on Wednesday, 28 July 2010 09:14:27 UTC