Re: The Spec: On the `edge' of despair

On Tue, 15 Nov 2011 11:38:59 -0700, Glenn Adams wrote:

> On Tue, Nov 15, 2011 at 11:13 AM, Tab Atkins Jr. <jackalmage@gmail.com>wrote:
> 
>>On Tue, Nov 15, 2011 at 7:22 AM, Michael Witten <mfwitten@gmail.com>
>>wrote:
>>> The CSS 2.1 specification uses some very confusing (if not erroneous)
>>> terminology in describing the fundamental concepts of the Box Model:
>>>
>>>  http://www.w3.org/TR/2011/REC-CSS2-20110607/box.html#box-dimensions
>>[snip]
>>> The uses of the words `perimeter', `edge', `width', and `box' are
>>> completely unreasonable unless the following statements are
>>> considered reasonable:
>>>
>>>    * The `perimeter' of an area really means the `outer perimeter'
>>>      of an area.
>>>
>>>      The word `edge' is a suitable synonym for the word `perimeter'.
>>
>>Yes, this seems reasonable.  The spec says exactly that.
> 
> I don't read these as synonymous.

Neither do I!

> The perimeter of an area is merely the boundary of that area.

Then you'll certainly like the fact that I introduce the term `boundary'
in my improved version.

> In the present context, all areas are closed areas. In contrast, an
> edge is a component (segment) of the perimeter.

Firstly, the spec says that the perimeter IS CALLED an edge (though,
your idea of `edge' referring to a segment of the perimeter is
also introduced, which is just doubly terrible).

Of course, you can make a distinction between:

  * The outer perimeter of an area.
  * The inner perimeter of an area (non-existant for the content area).

and then saying that an *outer perimeter* is some sort of a *segment*
that we call `an edge' (what my improved version calls `a boundary').

Note, though: You cannot reasonably use the unqualified word `perimiter'
to refer to both the inner and outer perimeters.

This careful use of terminology extends to a more careful (and correct)
explanation of how the various AREAS are defined in terms of the various
outer perimeters. See here:

  Subject: Re: Sloppiness of `box', `area', and `width'
  Message-ID: <a889626e05754a22839767923e556b7f-mfwitten@gmail.com>
  http://lists.w3.org/Archives/Public/www-style/2011Nov/0254.html

> Since we are dealing only with rectangular areas in the present context,

We're not though; only the content area is rectangular.

That's precisely my point! I'm trying to be very careful to stick very
closesly to the concepts that have been introduced.

> we have four (named) edges (per area), but we have only one perimeter.

The spec, however, EXPLICITLY introduces the word `edge' as a synonym
for perimeter, and first just IMPLICITLY mentions 'four edges' per 'edge'
before even talking about splitting an `edge' into... well... `edges'.
Gah!

Now, there's nothing wrong with saying "an edge can have edges", as Tab
said (I'm perfectly capable of managing that kind of abstraction), but
it's not a very wise choice for conveying these concepts.

See the bottom of this email:

  Subject: Re: The obvious confusion of `edge'
  Message-ID: <5a268ce5a67545818a85b38f67ef5715-mfwitten@gmail.com>
  http://lists.w3.org/Archives/Public/www-style/2011Nov/0253.html

> Note that it is useful to distinguish between
> 
> (1) absolute direction names of edges (left, right, top, bottom) [actually
> relative to what is (at some time) called left, right, top, bottom on the
> presenting media] and
> 
> (2) writing mode relative direction names of edges (start, end, before,
> after)

Yes, but that is not in question.

>>>    * The word `width' is a suitable synonym for the word `area'.
>>
>> No, that's not true, and it's not suggested in the text.  The "width"
>> talked about in this chapter is the distance between a box's edge and
>> the nearest enclosed box's edge. This can be different for each of the
>> four edges.  It's almost never synonymous with "area", as even when
>> the padding is zero, for example, the padding box will have the same
>> area as the content box.
>>
> 
> agreeing with Tab, width is clearly *not* synonymous with area
> 
> also, like the absolute edge names, width is an absolute extent name
> meaning the distance between left and right edges

See first email I reference above.

Received on Tuesday, 15 November 2011 20:06:59 UTC