W3C home > Mailing lists > Public > www-style@w3.org > July 2009

[CSSWG] Minutes and Resolutions 2009-07-22

From: fantasai <fantasai.lists@inkedblade.net>
Date: Wed, 22 Jul 2009 13:01:44 -0700
Message-ID: <4A677028.2080909@inkedblade.net>
To: www-style@w3.org

   - Discussed interaction of box-shadow and border-image. Chris presented
     his investigation and proposals in
     The discussion concluded that different types of border-images require
     different approaches. We will be looking for feedback on www-style.

   - Discussed percentage border-radii: elliptical vs circular, desired
     effects, and implementation status.

====== Full minutes below ======


   Bert Bos
   David Baron
   Arron Eicholz
   Elika Etemad
   Sylvain Galineau
   Daniel Glazman
   Brad Kemper
   Chris Lilley
   Peter Linss
   Steve Zilles

<RRSAgent> logging to http://www.w3.org/2009/07/22-CSS-irc
<ChrisL> Agenda: http://lists.w3.org/Archives/Member/w3c-css-wg/2009JulSep/0025.html

Scribe: ChrisL
Chair: Peter

Interaction of box-shadow and border-image

   ChrisL: I sent in a proposal
   <dbaron> I think we should post this proposal to www-style and try to get
            some feedback.
   Bert: Prefer the 4.2 opaque box solution
   Brad: prefer the one with shadows all round
   ChrisL: That's actually two shadows, one inset
   Daniel: Prefer the one with the alpha chanel.
   ... often asked how to do that
   Brad: Anyone that can create an image can also create a drop shadow
   Daniel: I have a counter example, I'm making an app and want the user agent
           to be able to use the alpha channel of a given image.
   dbaron: I think I might lean towards agreeing with BK, because this feels
           like the type of feature where we spend a lot of work implementing
           something that doesn't quite do what authors want, so nobody uses it.
   dbaron: Prefer to disable box-shadow
   ChrisL: I have an example not linked in which puts a drop shadow on the
           source; the slicing is then wrong
   dbaron: yes, can't get that with the author making their own shadow in the
           source image
   Daniel: Shadows not aligned with circles, so slicing will mess up the shadow.
           Would need two images, one for the shadow and one for the image itself
   SteveZ: If images distort the shadow distorts differently
   Daniel: Web designers want to apply the effects dynamically
   ChrisL: Could animate the drop shadow. Lighting effect on mouse position for
   <sgalineau> example of ChrisL's dynamic shadow update based on mouse position
               with text-shadow:
   <sgalineau> http://www.zachstronaut.com/lab/text-shadow-box/text-shadow-box.html
   Brad: Agree it's powerful, but should be a separate property that works on
         all images
   SteveZ: how about renaming the property border-shadow instead of box-shadow
   fantasai: everyone is using this property already, so its hard to change
             even with vendor prefixes
   dbaron: people will have to border-image change anyway
   ChrisL: Creating a version of the image with drop-shadow and then slicing
           and tiling it was really gross; snippets of shadow tiled across the
   fantasai: I think WebKit's problem with the border-image changes was that
             they're using it in applications in WebKit-only OSX environments,
             not just on the Web.
   fantasai: In these applications the spacing is more critical
   dbaron: I think the reason the change would cause more breakage for iPhone
           (etc.) applications than Web pages is that Web pages aren't using it
           because not all browsers support it yet
ScribeNick: fantasai
   Chris: I had a section in the document that shows the result with a solid box
   Chris: to show that it's the same as with drop-shadow on a solid border
   Brad: Already starting to think about what it currently does with box-shadow
   Brad: It doesn't follow the alpha-image of the actual border
   Brad: not using alpha channels
   CL: alpha-image of a solid rectangle is a solid rectangle
   Brad: But it's not if it's dots or dashes
   Brad: For this image seems like you can't slice it very well
   Brad: But for a lot of things I would use, e.g. for fancy corners on a
         straight-edged box
   Brad: would be able to use the images for the shadow
   Chris: I think when you actually try it, it won't work
   Peter: I'm sure there are some images where it would look right, but a lot
          where it wouldn't
   <sgalineau> can someone post the lamp demo page ?
   <fantasai> http://www.bradclicks.com/cssplay/border-image/Thinking_Outside_The_Box.html
   Brad: Either way you are going to be restricted
   Brad: E.g. in Chris's example you can't create complex effects
   Brad: Restriction on my way is that you can't animate it and certain types
         of images where you have a big corner and things narrowing as they
         come towards the corner.. that would be something you can't do with
         my way
   Steve: There's nothing that prevents someone from /not/ adding the
          drop-shadow and putting it in the image
   Brad: But then you can't use drop-shadow as a fallback
   fantasai: I think that's less important than being able to get these cases
   fantasai: We have a large chunk of use cases that can only be achieved by
             including the shadow in the image, and also a large chunk of use
             cases that can only be achieved by dynamically applying the border
   fantasai: Using box-shadow as a fallback only gives you marginally better
             results than not having it available, and I don't think that margin
             is as important as being able to handle both chunks of use cases
   fantasai: Also image fallback isn't specific to this feature. Maybe we could
             use media queries in the future to detect images being turned off.

   Peter: I think box-shadow and border-image are separate things and I don't
          think they should be combined like this.
   Peter: I think we should have a border-shadow property instead.
   Peter: Put a switch on it to get different behaviors.
   <ChrisL> so then the dashed borders would also behave like this?
   Steve: I like the solution that has border-shadow apply to both regular
          borders and image borders
   Steve: and box-shadow does what it does now
   <dbaron> Brad: ... where box-shadow is not drawn when there's a border-image.
   * Bert wonders when somebody will suggest border-light-sources,
          border-ambient-light, border-viewer-distance, border-reflectivity,
          border-transform, border-texture, border-backside-color...
   Chris: So are we going to move forward with border-shadow now, or reserve
          that for a future version?
   * fantasai use SVG filters
   * dbaron agrees with Bert that more general mechanisms would probably be
            good here
   Chris: Are we adding this now?
   Chris: Just copy the box-shadow property definition and tweak it
   Chris: spread for arbitrary images is not defined
   Brad: We also talked about having switches in the property, might delay CR
   <ChrisL> i'm hapy to add a border-shadow to the spec, and add an example
            that has a dashed border
   dbaron: We also need to get this discussion on www-style
   fantasai: I want to make sure roc and hyatt agree with whatever we decide
             to do here
   Steve: update Chris's proposal document to include border-shadow
   Brad: They can also read the minutes of this meeting, too, right?
   fantasai: Why not put the switch on box-shadow?
   fantasai: You wouldn't use both properties at the same time anyway
   ACTION: clilley to revise the border-image-and-box-shadow proposal to make a
           border-shadow proposal, them make public
   Steve: In your examples, the shadow didn't look like it was on the box, it
          looked like it was on the border
   Steve: I would prefer Chris write it up as a separate property and then
          suggest that it could be merged
   <ChrisL> I can revise the proposal and make it public for Friday

Publication Status for Flexbox and css3-images

   Bert: Will be officially published tomorrow
   Bert: longer answer is, I had to fix markup errors in the drafts, so next
         time you update do cvs update first

Percentage Corners

ScribeNick: ChrisL
   Topic: Percentage corners
   fantasai: Add this or defer? Already added in Mozilla and they will be
             removing the vendor prefix soon
   fantasai: I prefer to defer it as we decided earlier, but wondering if it's
             possible for Mozilla to remove at this point or if we have to
             define this so it doesn't conflict with whatever we spec later on
   dbaron: Should be possible to remove it
   dbaron: Just need to remember to remove it when we drop the -moz- prefix
   Peter: This has been implemented since the 90s
   dbaron: Don't recall seeing actual uses of it
   Peter: Was added to support things like MacOS buttons
   BK: Will the prefix form continue to be supported?
   dbaron: probably not
   Peter: What's the problem with having percentage values in there?
   fantasai: There are two differing interpretations of what a percentage means
   Brad: Suggested on www-style how to handle them
   ... one is based on width, ohter is based on the relevant side
   dbaron: Agree with Hakon and say they are always based on width or height
   ... if you want a particular shape you will use particular units
   Brad: How do you get ovals?
   * sgalineau would not be surprised if some authors filed a bug when they got an oval...
   dbaron: One radius always based on width the other based on height
   SteveZ: So its based on the border thickness?
   dbaron: No the size of the box
   <dbaron> so you'd get a circle with border-radius: 50%
   <dbaron> or an ellipse
   <dbaron> depending on the shape of the box
   Brad: If you want quarter-circle corners that are not ellipses, then you
         don't use percentages
   Brad: 10% height is different from 10% width
   Brad: So if you want quarter circle corners, you just can't use percentages?
   Peter: The only thing you couldn't get that way is a percentage-based curve
          that is always circular
   Peter: I kinda like Brad's idea that the percentage is always based on the
          width if you only specify one percentage
   fantasai: Normally if there are two values and you can drop one, it's
             duplicated, but does not give different behaviour
   ... so this is not like other shorthands
   SteveZ: Its duplicating the computed value, not the specified one
   Peter: Another way to get that would be to have a width unit. 0.5width
   Peter: Or make new units like a width unit. 0.5W
   Peter: width: 50% is the same as 0.5W
   ... could be introduced down the road
   fantasai: Apparently Opera also has percentages implemented
   Bert: Any other case where you want percentages, apart from elliptical boxes?
   <anne2> (if this is about borders, we might have removed those due to compat issues)
   <anne2> (supporting percentages there, that is)
   <fantasai> anne2, yes, this is about borders
   <anne2> (sites were using it expecting it not to work)
   <sgalineau> i believe i've seen border-radius used to make a circle
   Peter: Mac buttons with rounded ends irrespective of button length
   Bert gives an example of resizing his window so the box resizes. The size of
        the curve increases until it cuts throught the padding and overlaps the
   Bert: I don't want things like that.
   SteveZ: If I make the corner a constant size and make the box bigger the
           corner becomes more square. Don't want that
   fantasai: Anne says percent border radius might have been removed from Opera,
   SteveZ: not been in a released build anyway
   SteveZ: Like Brad and Peter's suggestion with the single value
   Peter: Not hearing consensus, out of time

<RRSAgent> I have made the request to generate http://www.w3.org/2009/07/22-CSS-minutes.html
<fantasai> dbaron, what do you think of the # of values solution?
<dbaron> fantasai, I think it's ugly but I can live with it.
Received on Wednesday, 22 July 2009 21:02:19 UTC

This archive was generated by hypermail 2.3.1 : Monday, 2 May 2016 14:38:27 UTC