W3C home > Mailing lists > Public > www-style@w3.org > May 2013

[css-device-adapt] Zero sized viewport lengths and zoom values

From: Rune Lillesveen <rune@opera.com>
Date: Thu, 02 May 2013 13:43:54 +0200
Cc: "Simon Pieters" <simonp@opera.com>
To: "www-style list" <www-style@w3.org>
Message-ID: <op.wwgf7gx98isf1p@rune-optiplex-980>
Hi.

There are two related issues about zero widths/heights in @viewport  
reported by me[1] and Simon Pieters[2].

I'd like to resolve these issues by allowing zero lengths for  
<viewport-length>s, both for descriptor values and resolved/used values.

Rationale:

1. It's already possible to get zero-length widths and heights for  
viewports established by iframes.

2. Range checking[3] for calc() requires a closed interval value range.  
The current range is open.

Realizing that even initial viewports can have 0 lengths for width and  
height (in particular through iframes), means the spec should say  
something about potential division-by-zero in the constraining procedure.

Also, for the sake of range checking of calc(), I'd like to change to  
allow 0 values for the various *zoom descriptor values. In practice, a UA  
will have a lower limit on the used value for zoom, so I don't think this  
change will mean anything in practice.


Proposed changes:

1. Change all requirements for lengths in descriptors from "positive" to  
"non-negative".

2. Change all requirements for zoom factors in descriptors from "positive"  
to "non-negative".

To avoid division-by-zero problems in the constraining procedure:

3. Change:

   "7. If width is ‘auto’, set width = height * (initial-width /  
initial-height)"

   to:

   "7. If width is ‘auto’, set width = height * (initial-width /  
initial-height), or width = initial-width if initial-height is 0."

   and change:

   "8. If height is ‘auto’, set height = width * (initial-height /  
initial-width)"

   to:

   "8. If height is ‘auto’, set height = width * (initial-height /  
initial-width), or height = initial-height if initial-width is 0."


Issue [2] is about my test-suite checking for non-zero used values of  
viewport lengths. That can be resolved by me changing the test-suite.


[1] https://www.w3.org/Bugs/Public/show_bug.cgi?id=21782
[2] https://www.w3.org/Bugs/Public/show_bug.cgi?id=20289
[3] http://dev.w3.org/csswg/css-values/#calc-range

-- 
Rune Lillesveen
Received on Thursday, 2 May 2013 11:44:28 UTC

This archive was generated by hypermail 2.3.1 : Thursday, 2 May 2013 11:44:28 UTC