Re: Level of specification detail

At 3:00 PM -0700 UTC, on 4/11/07, Maciej Stachowiak wrote:

> On Apr 11, 2007, at 8:24 AM, Dan Connolly wrote:

[...]

>>> On Wed, 11 Apr 2007 11:28:36 +0200, Henrik Dvergsdal
>>> <henrik.dvergsdal@hibo.no> wrote:

[...]

>>>> In most cases authors/developers don't need the algorithms - they
>>>> just need to validate the syntax and then check with the browsers
>>>> if things work/look ok.

Web publishers also need to understand how search engines deal with what they
publish. Most web publishers *cannot* check with browsers to see if "things
work OK".  There are too many possible different browsing environments to
check.

[...]

>> There is a natural tension between writing the sort of precise,
>> exhaustive specification that meets the needs of implementors,
>> and writing for the audience of authors

Indeed.

[...]

> The spec itself should mainly try to address implementors of user
> agents (including browsers, search engines, data mining tools, etc),
> conformance checkers and authoring tools, since these audiences need
> a significant amount of precision.

If the spec is to positively affect the Web, it will need to be written such
that it can be easily understood by web publishers[*]. Every effort must be
made to make the spec clear and understandable for them. Leaving this up to
third parties will result in continuation of today's 99% incomplete and
misguided tutorials.

Targeting different groups (UA authors, web publishers, etc.) with the same
text is extremely difficult, so it may be that the spec will need to consist
of different versions, targeted specifically at those groups. (Which isn't
easy either -- the risk of contradiction.)


[*] on the WHAT WG mailing list it has often been argued that authoring tools
in fact produce "hand coded" HTML as well, so making the spec understandable
for authoring tools authors would implicitly be the same requirement as
making it understandable for web publishers.


-- 
Sander Tekelenburg
The Web Repair Initiative: <http://webrepair.org/>

Received on Thursday, 12 April 2007 15:41:36 UTC