Re: ISSUE-140 CPP — no conformance versions

2011-02-06 14:19, Anne van Kesteren skrev:
> Summary: Conformance to HTML should not have versions.

Answering Anne, but I've read the whole thread up to now.

I can live with the dual mode of WhatWG going versionless while the W3C 
has versions. But as a teacher I would find it very hard to not have a 
fixed profile when using a conformance checker. Making ad hoc profiles 
as suggested by Benjamin H-L could be an opt-in, but especially for 
newbies advice like the following will just add extra confusion:

"Use markup patterns described in documents L, M, N.
Validate using tool X using options Q and R. Ignore errors A, B, C.
Lint using tool Y. Ignore warnings G, H, I."

For quite some time now, the doctype has been the contract to what 
flavour of (X)HTML a developer is using, and it has worked fairly well, 
even though some rules have been de facto bad (e.g. disallowing 
numbering on numbered lists). Validation according to fixed set of rules 
works mentally. Validation performed against a floating target does not.

Putting my teacher's hat on:

One can not expect teachers and students to keep track of every new 
feature or change in HTML, as soon as it has been put into the spec. If 
it did validate on Monday when the assignment was turned in, it should 
still validate on Friday when it is being graded. I think this can be an 
issue for companies dealing with clients as well. "But you said you 
would deliver valid code..."

Perhaps a reasonable compromise could be to have conformance checkers do 
"profiles", instead of versions, like JSLint having a pre made setting 
called "the good parts". But that would of course lead us down the path 
of having a plethora of profiles and until this horse is out of the 
stable, lets KISS!

As for conformance versions giving a wrong "impression", speaking as a 
teacher, I think that is a far lesser problem to deal with.

Another thing to consider: Getting an all green from the validator is a 
boost of morale for young students. Simple, but I've seen it again and 
again. Whenever I encounter a situation in the CSS validator where there 
is a bug and the green is unavailable for the wrong reasons, there may 
be an intellectual acceptance of the fact, but it does not work as a pat 
on the back with the same force as getting a real "green" does.

Of course validation against an obsolete or buggy set of rules might 
produce a false sense of security, but once again, as a teacher my 
experience is that this also is a lesser problem.


-- 
Lars Gunther
http://keryx.se/
http://twitter.com/itpastorn/
http://itpastorn.blogspot.com/

Received on Sunday, 6 February 2011 22:59:02 UTC