W3C home > Mailing lists > Public > www-style@w3.org > May 2014

Re: [css-flexbox] Better define the computed value of 'flex-basis'

From: fantasai <fantasai.lists@inkedblade.net>
Date: Mon, 19 May 2014 05:52:52 -0700
Message-ID: <5379FEA4.1030903@inkedblade.net>
To: www-style@w3.org
On 05/07/2014 11:57 AM, fantasai wrote:
> On 05/07/2014 11:16 AM, Tab Atkins Jr. wrote:
>> Right now 'flex-basis' has its computed value defined as "as
>> specified, with lengths made absolute".  This implies that
>> "flex-basis: auto" (which means "use the value of width/height") stays
>> as "auto" in the computed value, and it's not well-defined when this
>> changes into the width/height value.
>>
>> fantasai and I propose that we define that "flex-basis: auto" computes
>> to the width/height value.  Rossen concurred in a private discussion.

So, this was discussed in the WG today, and because it means
that a computed value of 'auto' cannot not compute to itself,
and exposes some details of exactly how inheritance and cascading
work in the engine, the proposal to make 'auto' compute through
to the 'width'/'height' value was rejected.

We still have the problems mentioned, however, plus another one
I just noticed: there is no way to explicitly request a flex
basis of 'auto', other than to set width/height also to auto.
In the past, 'auto' was equivalent to 'max-content'. But since
then we added changes to handle intrinsic aspect ratios and
various layout algorithm rules now depend on whether the flex
basis is auto.

It might be possible to fix this by renaming 'flex-basis: auto'
to something else (while keeping it as the initial value) and
retaining the current behavior of 'flex: auto'. Something to
consider.

~fantasai
Received on Monday, 19 May 2014 12:53:22 UTC

This archive was generated by hypermail 2.3.1 : Monday, 19 May 2014 12:53:22 UTC