Re: [css3-images] Probably editorial: 'object-fit' values 'cover' and 'contain' contain redundant definition

Tab Atkins Jr. <> skreiv Wed, 22 Feb 2012 03:06:48  

> On Thu, Feb 2, 2012 at 4:12 AM, Tab Atkins Jr. <>  
> wrote:
>> On Wed, Feb 1, 2012 at 7:07 AM, Leif Arne Storset <>  
>> wrote:
>>> In section 5.4, both 'contain' and 'cover' seem to be defined twice.  
>>> For
>>> example, 'contain' says
>>> | Determine the used ‘height’ and ‘width’ of the element as usual, |  
>>> except: If both ‘height’ and ‘width’ are ‘auto’, and the used
>>> | value of at least one of ‘max-width’ and ‘max-height’ is not
>>> | ‘none’, then compute the element's used width and used height as
>>> | though the intrinsic dimensions of the contents were infinitely
>>> | large numbers whose ratio is the actual intrinsic ratio of the
>>> | contents. This will proportionally scale the used width and
>>> | height up to the given maximum constraints.
>>> |
>>> | Set the concrete object size to the largest width and height that has
>>> | the same aspect ratio as the object's intrinsic aspect ratio, and
>>> | additionally has neither width nor height larger than the replaced
>>> | element's used width and height, respectively.
>>> These seem redundant to me. I suspect the first definition, written in
>>> 2007, hails from a time when '(max|min)-(width|height)' were  
>>> underdefined
>>> in level 2, and "as usual" wasn't sufficient. [0] Determining used  
>>> values
>>> for 'width' and 'height' seems out of scope for this property and this
>>> spec, as implied by the introduction to 5.4:
>>> | The ‘object-fit’ property specifies how the contents of a replaced
>>> | element should be scaled relative to the box established by its used
>>> | height and width.
>>> I could be wrong: my head is spinning from trying to calculate
>>> some examples according to both the specs. If I am wrong, and some
>>> situation does require all the infinity math, I feel we could phrase it
>>> more simply and/or state the covered cases more clearly.
>>> My primary suggestion, however, is to delete the first paragraph, both
>>> under 'contain' and 'cover', so that both consist only of a paragraph
>>> starting with "Set the concrete object size…".
>>> [0] At the moment I can't find old versions of CSS 2.1, but CSS 2 does  
>>> not
>>> define it as well as CSS 2.1.
>>> CSS 2:
>>> CSS 2.1:
>> Nope, it's not redundant.  Those first paragraphs have the effect of
>> forcing the size of the element itself to the min/max constraint.
>> I agree that this is out-of-scope for this property.  Fantasai does not.
> Since I rejected this feedback, could you indicate whether you're okay
> with this resolution or not, Leif?  This is necessary for the
> Disposition of Comments.

Ah, didn't realize that was a rejection (rather than a postponement).

I disagree (i. e. agree that it is out of scope). You're right that they  
are not redundant with each other, but the first definition is redundant  
with the now more robust CSS 2.1 section 10.4.

I can live with it, though, since it's not incorrect, and I've already  
implemented and understand what it's talking about. :) I do think it is  
very confusing for first-time readers.

(Just now I realized that the introduction also mentions this behavior:  
"[The property] also enables scaling a replaced element up to a specified  
maximum size or down to a specified minimum size while preserving its  
aspect ratio.". That should also have been deleted in my change proposal.)

If you do keep it (and it's not too late in the process for editorial  
changes!), I would suggest adding a reference to CSS 2.1 section 10.4,  
where element sizing is defined more explicitly. That way, first-time  
readers will get that this part of the definition deals with something  
different than the other part. Such as:

| 'contain'
| This will proportionally scale the used width and height up to the
| given maximum constraints.
+ (See [CSS21] section 10.4 for more information on min/max constraints.)
| Set the concrete object size to the largest width and height that has

and similarly for 'cover'.

Leif Arne Storset
Core Technology Developer, Opera Software
Oslo, Norway

Received on Wednesday, 22 February 2012 11:00:48 UTC