On Wed, Nov 5, 2014 at 12:06 AM, Bruno Racineux <bruno@hexanet.net> wrote: > On 11/3/14 1:45 PM, "Daniel Holbert" <dholbert@mozilla.com> wrote: >>The "main-size" rename has only made it as far as Firefox's "beta" >>release -- it hasn't made it to an official release yet. It's possible >>we'd discover more breakage on the release channel. > > Why do implementors have to break the previous behavior in the first place? > > Can't browsers implement the new keyword and keep the old 'auto' keyword > for a while, for backward compatibility reasons, and just issue warnings > in the console to avoid said breakage if proven a significant problem? To resummarize: Previousy, the 'flex-basis' property (accidentally) had two completely different "auto" values; an "auto" in a *specified* value meant "use the value of width/height", while "auto" in a *computed* value meant "auto-size". This was dumb and confusing, so once we realized it, we split the two meanings apart. We had two choices of what to do: 1. Keep "auto => use width/height", and add a new keyword "content" for "auto-size". 2. Keep "auto => auto-size", and add a new keyword "main-size" for "use width/height". #1 is good, because it keeps "auto" consistent with current behavior. #2 is good, because it means you can use any width/height value in flex-basis and have it mean the same thing. We went with #2, because it's what we would have chosen in the first place had we realized we were doing something confusing. We added a few wrinkles (using "auto" as the sole value in the 'flex' shorthand still triggers the "use width/height" behavior) to help with compat, and hope that we can still keep it, but it's looking chancy. We may end up having to do #1 just for compat reasons, unfortunately. > PS: The spec still refer to 'flex-basis:auto' on the svg Figure 6. Thanks, will fix. ~TJReceived on Wednesday, 12 November 2014 00:42:28 UTC
This archive was generated by hypermail 2.4.0 : Monday, 23 January 2023 02:14:45 UTC