W3C home > Mailing lists > Public > www-style@w3.org > December 2010

Re: [CSS21] 4.3.2 Lengths (reference pixel?)

From: Dr. Olaf Hoffmann <Dr.O.Hoffmann@gmx.de>
Date: Fri, 10 Dec 2010 18:34:06 +0100
To: www-style@w3.org
Message-Id: <201012101834.07465.Dr.O.Hoffmann@gmx.de>
Tab Atkins Jr.:
>
> I suggest rereading the previous discussions on the matter.  This
> subject has been gone over ad nauseum.  In short, no, you don't always
> know what a mm is to any great precision.  That depends on the output
> device.

Well, you don't have to explain me my job. Precision is always limited to
some error margin. If I measure a length (or something else), I have a good 
guess about the precision of the measurement. 
A proper measurement contains an error margin. And if it is not good enough
for me, it is my task to improve the tools I use. Therefore for my purposes,
I know what a mm is with a sufficient precision, if I really want to know it.
Other people in my institute for example have the capability to measure
length changes significantly smaller that an atom nucleus ;o)

And with relation to W3C works, the SVG recommendations show, how
this can work. They contain roughly the requirement that the presentation 
accuracy has to be at least 1 device pixel. Therefore, with this precision
everyone can test, whether 10cm is presented as 10cm or not. 
The CSS group could add something like this to the CSS draft to get 
something testable.

But back to the issue:
I comment on what is written/changed now in the new draft - my assumption
is, that this is somehow a conclusion of discussions within the CSS WG - it
is incompatible with all previous CSS recommendations and drafts and
it contains and inconsistency in itself. However incompatibilities with
other recommendations or drafts are another issue, we should not mix
up - this can be solved, after the current draft can be considered to be
at least consistent.
And if this is the result of a previous discussion, we have to discuss the
problems of this result and not necessarily, on which way the CSS WG 
or the editor of the draft draw conclusions from previous discussions.
And even more, if this is the result of a longer discussion, no matter 
what was discussed or what the participant think, the result of the
discussion was, what counts finally is what is written in the draft,
and this need obviously some fix, because it contains contradictions.
Sometimes it is that simple - if you find a contradiction, its history
does not matter anymore, it just has to be fixed.
I can propose fixes, but to do it - not just to discuss it - is the task
of the CSS WG here.



David Singer:

> > I think, lengths are only relevant for some kind of graphical
> > presentation, not for example for aural style sheets, removed anyway in
> > CSS2.1.
> No, this seems to be one of the areas where you misunderstand.  

Maybe the problem of understanding is on your side?
If the presentation contains no lengths, there is no method to apply length
values.

> Consider, 
> for example, projection glasses (which present an image directly to the
> eyes) -- there is no 'display surface' on which to measure.

Well, if you are looking for lengths within a projection, you can measure
or calculate it in the projection plane (it is more or less classical optics,
therefore simple).
At least my eyes have a macroscopic size, 
therefore everything is measurable (and predictable) in principle here.
Obviously authors need to provide a specific stylesheet for such
an application, if they insist here to use absolute units like mm.
I think, typically it is more clever not to use absolute units for such
an application - on the other hand, if the author knows precisely
the device and the intended application, it can be a good idea
to optimise everything with absolute units.

But as already mentioned, this is another issue, not related to
the inconsistency I commented on.
To avoid the inconsistency is the key to allow authors to
provide device specific stylesheets with absolute units,
if they think, this matters. With CSS2.0 they can, why not
with CSS2.1 (respectively why to scale px with absolute
units, if authors need device specific stylesheets based
on device pixels?).

> In other 
> cases, there is a display surface, but it's inappropriate to apply the
> measurements directly (e.g. on a billboard display a single pixel may
> exceed a physical inch on the display surface).

Why is it inappropriate, if the author wants it?
If the author provides a stylesheet for such a device, maybe it
will contain information like font-size: 100cm
(and in some cases it may have some use, if the authors
knows exactly the device, the stylesheet is intended for).

But as already mentioned, this is another issue, not related to
the inconsistency I commented on.
>
> As others have indicated, please read the (extensive) threads on this
> subject, where this was discussed.

See above.

This comment about the current draft ist not about previous discussions,
it is about the current draft, it is dated from 2010-12-07 and there is no
previous discussion about this section of the draft related to the indicated
inconsistency.  If you read it, have a look on the dates and the mailing list,
you will easily see this. Because the inconsistency is still in the draft, 
obviously if there was already an internal discusssion of the CSS WG, it
did not solve the problem, else it would not be in the draft ;o)
Respectively give me a link to a discussion related to this issue about
the draft from 2010-12-07, if I am wrong - better give me a link to
the solution of the problem of the contradiction, if the problem is already
solved after 2010-12-07.

Olaf
Received on Friday, 10 December 2010 17:34:43 GMT

This archive was generated by hypermail 2.3.1 : Tuesday, 26 March 2013 17:20:35 GMT