Versioning and error handling (was Re: Semicolon after entities)

On Thu, Apr 26, 2007 at 11:49:01AM +0100, Nicholas Shanks wrote:
> "Sorry, the website at web.example.com could not be displayed because  
> the developers are incompetent.
> 
> Error on line 30:  invalid type value ?combobox? for input element."

This could be a problem. Say that HTML 5 didn't have a combobox type,
but HTML 6 did. Refusing to render it at all wouldn't be a great idea.

... of course, if we had versioning then a more liberal mode could be
adopted for version numbers in the future. (hmm, this document is
written in an unsupported version of HTML, attempting to render anyway, IE9
supports this version [upgrade now]).

The other issue is that mistakes DO happen. OTOH, QA tools do exist
and adopting a publishing process that involves the use is a QA tool
isn't all that difficult.

> I believe versioning of documents is superior to non-versioned  
> document formats, as does the overwhelming majority of the computer  
> industry.

One argument that I haven't seen mentioned is QA testing. How is a QA
tool supposed to test a document for conformance to a specification if
the document doesn't indicate what specification it is written against?

-- 
David Dorward                                      http://dorward.me.uk

Received on Thursday, 26 April 2007 10:58:05 UTC