[skew/shear] Re: Minutes, FX telcon 11 March 2010

On Thu, Mar 11, 2010 at 4:28 PM, Chris Lilley <chris@w3.org> wrote:
> Hello public-fx,
>
> Minutes are here
> http://www.w3.org/2010/03/11-fx-minutes.html
>
> and below as text for bots
>
>                   CSS-SVG Task Force Teleconference
>
> 11 Mar 2010
>
>   [2]Agenda
>
>      [2] http://lists.w3.org/Archives/Public/www-svg/2010Mar/0007.html
>
>   See also: [3]IRC log
>
>      [3] http://www.w3.org/2010/03/11-fx-irc
>
> Attendees
>
>   Present
>          ed, Doug_Schepers, ChrisL, anthony, +1.858.655.aaaa, plinss_,
>          +1.408.454.aabb, dino, smfr, fantasai
>
>   Regrets
>   Chair
>          Erik
>
>   Scribe
>          Chris
>
> Contents
>
>     * [4]Topics
>         1. [5]Charter and so on
>         2. [6]CSS 2d Transforms and SVG Transforms differences
>     * [7]Summary of Action Items
>     _________________________________________________________
>
>   <trackbot> Date: 11 March 2010
>
>   <ChrisL> trackbot, start telcon
>
>   <trackbot> Meeting: CSS-SVG Task Force Teleconference
>
>   <trackbot> Date: 11 March 2010
>
>   <ChrisL> zakim - code?
>
>   <shepazu> code is FXTF
>
>   <ChrisL> hi dino
>
>   <dino> hmm
>
>   <shepazu> dino!
>
>   <dino> duh - wrong number me!
>
>   <anne> fwiw, regarding the suggestion about having some kind of auto
>   value
>
>   <anne> I don't think that is needed at all
>
>   <anne> see [8]http://krijnhoetmer.nl/irc-logs/css/20100311#l-61
>
>      [8] http://krijnhoetmer.nl/irc-logs/css/20100311#l-61
>
>   <shepazu> anne, will you join the call?
>
>   <shepazu> [9]http://www.w3.org/Graphics/fx/charter/
>
>      [9] http://www.w3.org/Graphics/fx/charter/
>
>   <anne> shepazu, no sorry, don't really know much about transforms
>
>   <shepazu> [10]http://www.w3.org/Graphics/fx/track/
>
>     [10] http://www.w3.org/Graphics/fx/track/
>
>   <anne> also, I should really fetch some dinner
>
>   <ChrisL> Scribe: Chris
>
>   <ChrisL> ScribeNick: ChrisL
>
> Charter and so on
>
>   Erik: Any change requests for the charter?
>
>   Doug: Clarify what font/textwrap means
>
>   Chris: Optical alignment
>   ... no longer there, corrected to "Wrapping text to a shape"
>
>   Doug: Need to update the participant list
>
>   Erik: UTF-8 issues, d'oh!
>
>   Simon: Dean is my hero
>
>   <shepazu> [11]http://www.w3.org/Graphics/fx/track/
>
>     [11] http://www.w3.org/Graphics/fx/track/
>
>   Doug: Other issue is whether we have a lot of small specs or a
>   smaller number that cover both SVG and CSS
>   ... most seem in favour of specs that cover both
>
>   Erik: fine by me
>
>   Simon: Would joint deliverables approve all publications?
>
>   Doug: Yes
>
>   Simon: concerned over publication delays
>   ... would this replace or be in addition to the CSS transforms spec?
>
>   Doug: Replace
>
>   Simon: Agree in principle
>
>   Chris: Harmonisation takes longer than agreeing on joint publication
>
>   Doug: Reasonable concern but manageable, deal with it if it arises
>   ... we like putting trasform-x, transform-y on their own attrs.
>   Don't recall any blocking iissues in the spec
>   ... want to make sure it applies equally to CSS/HTML and SVG
>
>   Simon: OK sounds good, enough representation from both groups
>
>   Dino: Point is to take it out of scope of one group?
>
>   Chris: no, TF work has to be in scope for all WGs that contribute
>
>   Dino: So issue is really just about status, naming and so on
>
>   Erik: So keep it in CSS cvs space?
>
>   Dino: Don't care where it lives
>
>   Doug: Maybe easier to re-use, or to make a new one. WebApps reuses a
>   repository from 2 groups, not a problem in practice as longas its
>   documented
>
>   Anthony: Don't care as long as its consistent.
>   ... prefer all TF docs in one place
>
>   <scribe> ACTION: Doug to create an FX repository [recorded in
>   [12]http://www.w3.org/2010/03/11-fx-minutes.html#action01]
>
>   <trackbot> Created ACTION-1 - Create an FX repository [on Doug
>   Schepers - due 2010-03-18].
>
>   Resolved: We will have one FX TF repository
>
>   Doug: Editors - Dino and Anthony?
>
>   Anthony: Love to
>
>   Dino: Yippeee
>   ... Current spec has multiple editors, Simon isn't but should be,
>   Hyatt is but should not, DBaron is an active editor too
>
>   Doug: Suggest folding in all editors
>
>   Dino; Actually not sure on Dbaron, in terms of editong, but he is
>   certainly implementing
>
>   Resolved: Editors for Transforms will be Dino, Simon, Anthony, and
>   DBaron welcome too if he edits
>
> CSS 2d Transforms and SVG Transforms differences
>
>   Erik: sent an email about how to implement CSS transforms in SVG
>
>   <ed>
>   [13]http://lists.w3.org/Archives/Public/www-svg/2010Mar/0013.html
>
>     [13] http://lists.w3.org/Archives/Public/www-svg/2010Mar/0013.html
>
>   <ed> annevk had some comments on the auto-suggestion,
>   [14]http://krijnhoetmer.nl/irc-logs/css/20100311#l-61
>
>     [14] http://krijnhoetmer.nl/irc-logs/css/20100311#l-61
>
>   Erik: need to maintain backwards compat, and properties et in style
>   sheets have higher specificity than any value set in a attribute
>   ... Most comments seem to agree with the option 1 in my mail
>
>   Dino: Trying to understand what point Anne makes, the initial value
>   should use the attribute
>   ... doesn't quite work because transitions or animations that go
>   through 'none' should go to the null transform
>
>   Chris; So we need separate values for identity transform and "use
>   the attribute"
>
>   Dino: Yes
>
>   Simon: Exactly. Any transform creates a CSS stacking context and a
>   positioning container
>
>   Doug: Also times when ther e is a transform that you want to blow
>   away using CSS so we need that capability
>
>   Dino: Keyword "identity"?
>   ... especially for animations, using lists. Transforms in a list
>   animated separately, to avoid flattening rotations > 360 degrees.
>   Should add an identity keyword. Still need an auto as an initial
>   value, like none in CSS and does nothing in SVG
>   ... identity is still a transform
>
>   Anthony; Can none be replaced by identity there
>
>   Dino: yes
>
>   Chris: So identity would deal with the 'blow off transform' case?
>
>   Dino: Yes
>   ... Rotate 720 when flattened gives rotate 0 so no movement instead
>   of spinning twice
>
>   Anthony: Makes sense. Carry across to SVG as well
>
>   Dino: SVG does not cover this and it should
>   ... CSS3 units spec requires normalisation to 0..360 and that has to
>   be removed
>
>   <smfr>
>   [15]http://www.w3.org/TR/2009/WD-css3-2d-transforms-20090320/#animat
>   ion
>
>     [15] http://www.w3.org/TR/2009/WD-css3-2d-transforms-20090320/#animation
>
>   Chris: Normalisation came from spatialised audio (where it made
>   sense) but fpr geometric transformation, especially when animated,
>   notrmalisation should be avoided
>
>   Simon: CSS transitions applying to SVG, intent to have these work
>   also on transforms in SVG?
>
>   Doug: Authors would expect that
>
>   Simon: aaargh (scribe missed complication)
>
>   Dino: Style sheet trumps attribute, transform as a property also
>   trumps it. But SMIL trumps the attributes
>
>   <smfr> css transitions/animations are animating css transforms in
>   svg, while at the same time svg animations may be animating the
>   transform attributes
>
>   Erik: Use end events for transitions. Complex when they are both
>   animating the same property
>
>   Chris: Animation sandwich should define that
>
>   Dino: Yes but if there is an animation, to or by so going from
>   current value, and a transition fires ....
>
>   Chris: Current means current animated value. need to say whether a
>   transition is an animation in that sense
>
>   Doug: Jack Jansen interested to simplify SMIL animation to better
>   fit CSS
>
>   Dino: Suggest we leave that for now as its more about trnsitions
>   than transforms
>   ... Transitions work really well with style changes, so making
>   transforms as style helps there
>
>   Simon: Specifying behavior is tricky to work between smil animation
>   and css animation. One solution is to map all css transforms down to
>   transforms in attributes for example. So its all mapped to one value
>   space
>
>   Anthony: Agree this needs to be considered, for implementors and
>   authors, ease of authoring and also performance
>
>   <fantasai> RRSAgent: pointer
>
>   Simon: Way we treat historical html attrs like bold etc
>
>   [16]http://dev.w3.org/SVG/profiles/1.1F2/publish/styling.html#UsingP
>   resentationAttributes
>
>     [16] http://dev.w3.org/SVG/profiles/1.1F2/publish/styling.html#UsingPresentationAttributes
>
>   For user agents that support CSS, the presentation attributes must
>   be translated to corresponding CSS style rules according to rules
>   described in Precedence of non-CSS presentational hints ([CSS2],
>   section 6.4.4), with the additional clarification that the
>   presentation attributes are conceptually inserted into a new author
>   style sheet which is the first in the author style sheet collection.
>   The presentation attributes thus will participate in the CSS2
>   cascade as
>
>   <fantasai> um
>
>   <fantasai> that got cut off
>
>   <fantasai> also
>
>   Erik: So we would need to move the SVG transform attribute to the
>   CSS transform property. Mostly this works but in some cases not
>
>   <fantasai> the rule *is* that the attributes are inserted into a new
>   author style sheet that is the first in the author style sheet
>   collection
>
>   Erik: ref transform in SVG 1.2 is an example'Simon: right
>
>   <fantasai> and that they have zero specificity
>
>   <fantasai> so that all subsequent rules override the attributes
>
>   <fantasai> that doesn't require any "additional" clarification, it's
>   in the spec
>
>   Simon: So that aligns with Erik's option 1 in his email
>
>   Erik: yes
>
>   <fantasai> SVG just needs to say that its attribues are handled as
>   non-CSS presentational hints like <font face>
>
>   <fantasai> also, wrt 360 normalization, I think we already decided
>   to fix that. You just need to pester howcome about editing it into
>   the draft
>
>   <fantasai> or get someone else to edit it
>
>   <shepazu> fantasai, can you join the call?
>
>   <fantasai> passcode?
>
>   <smfr> FXTF (3983)
>
>   ok so 858 is Peter and 408 is Anthony. Moving on .....
>
>   Doug: (explains recap on hints)
>
>   Chris (explains re legacy, hints, 1:1 mapping)
>
>   Elika: CSS 2.1 or CSS3 cascading can be tweaked if needed. no
>   aditional clarification in SVG is needed
>
>   <fantasai> "For other languages, all document language-based styling
>   should be handled in the user agent style sheet. "
>
>   Elika: which is wrong
>   ... That spec should be made more SVG freindly
>
>   Chris: I agree and think the wording in SVG is a condensation of my
>   various explanations of how CSS works
>
>   Erik: So we agree on mapping SVG transform attr into the
>   corresponding CSS property. So we need an explicit mapping -
>   volunteers?
>   ... rotate function with three params in svg, one in CSS because of
>   transform-origin property
>
>   Dino: Or we could add rotate with three params. Try to avoid
>   argument overloading
>
>   Anthony: By default rotates happen around centre of object, so does
>   two extra values give you an offset?
>
>   Dino: Form the current transform origin
>
>   Chris (problems with current point in PostScript)
>
>   Simon: transform is a pre-shift and a post shift
>
>   Dino: Easty to add it
>   ... temporary translate
>
>   Erik: Commas, units (not allowed in SVG). Are units mandatory?
>
>   Dino: Yes, except scale obviously. But yes for lengths and angles
>   ... degrees default for angles
>   ... So want to not require commas?
>
>   Erik: Yes, not an issue really
>
>   Simon: Happy to drop requirement for commas in CSS
>
>   Dino: Skew?
>   ... We hate skew!
>
>   <ed>
>   [17]http://dev.w3.org/SVG/profiles/1.1F2/publish/coords.html#Transfo
>   rmAttribute
>
>     [17] http://dev.w3.org/SVG/profiles/1.1F2/publish/coords.html#TransformAttribute
>
>   Simon: CSS happy to drop skew , or skew-x, skew-y
>   ... skew with 2 params. Happy to move to what SVG does there with
>   separate skew-x, skew-y
>
>   Dino: Animating skew tends to look wierd if it uses matrix
>
>   Resolved: Remove skew, add skew-x and skew-y
>
>   Erik: ref transform
>
>   <ed> [18]http://www.w3.org/TR/SVGTiny12/coords.html#transform-ref
>
>     [18] http://www.w3.org/TR/SVGTiny12/coords.html#transform-ref
>
>   Chris: Ref lets you unwind the CTM to a different element
>
>   Simon: Inverse transform may be awkward on 3D if they dont share a
>   common 3d space
>
>   Chris: Same if you have HTML with two SVG subtrees with different
>   viewboxes. no common world coordinate system
>
>   Erik; Can only specigy svg in the syntax for the CTM
>
>   <shepazu> scribeNick: shepazu
>
>   dino: I don't see why you need transformRef in the first 2
>   examples...
>
>   ed: it's for moving back to something you can control, as opposed to
>   zooming and panning
>
>   simon: position fixed
>
>   dino: we could put it in, but have it not apply, or have it as a
>   translate...
>
>   ed: not sure it makes sense in CSS
>   ... it would have to take precedence in some way
>
>   dino: is this used and useful?
>
>   ed: our implementation is slightly different than the spec
>
>   <ChrisL> ScribeNick: ChrisL
>
>   <shepazu> Issue: consider not adding transformRef to Transforms spec
>
>   <trackbot> Created ISSUE-1 - Consider not adding transformRef to
>   Transforms spec ; please complete additional details at
>   [19]http://www.w3.org/Graphics/fx/track/issues/1/edit .
>
>     [19] http://www.w3.org/Graphics/fx/track/issues/1/edit
>
>   <scribe> ACTION: Simon to remove skew and add skew-x and skew-y to
>   consolidated transform spec [recorded in
>   [20]http://www.w3.org/2010/03/11-fx-minutes.html#action02]
>
>   <trackbot> Sorry, couldn't find user - Simon
>
>   trackbot, status?
>
>   <smfr> darn :)
>
>   <scribe> ACTION: Chris to Simon to remove skew and add skew-x and
>   skew-y to consolidated transform spec [recorded in
>   [21]http://www.w3.org/2010/03/11-fx-minutes.html#action03]
>
>   <trackbot> Created ACTION-2 - Simon to remove skew and add skew-x
>   and skew-y to consolidated transform spec [on Chris Lilley - due
>   2010-03-18].
>
>   <dino> my suggestion was that transform-ref should have been a
>   separate property/attribute, rather than part of the transform
>
>   Chris: in retrospect I agree
>
>   Doug: Some parts of SVG Tiny 1.2 could be rethought in SVG 2.0
>
>   <dino> for example, what does transform="ref(svg, 100, 100) ....
>   ref(svg, 200, 200) ... " mean?
>
>   <dino> anyway.
>
>   Erik; Although its useful, and making it available through CSS
>   transforms would be fine
>
>   Chris: Do we have merged specs or do we need to check in both and
>   then work on a third?
>
>   Anthony; Have started a merge for transforms, but its harder than it
>   looks
>
>   Chris: Best to have archival copies to refer to during the merge?
>
>   Dino: We currently use Bert's script to take HTML and make it
>   pubrules compliant. Also need to redirect shortnames. Publish with
>   new refs as soon as possible
>
>   Simon: And uodate CSS current work
>
>   Doug: And SVG wiki likewise
>
>   Simon: Need to know when the public FX one is the master so we can
>   get edits done
>
>   <shepazu> ACTION: Simon to remove skew and add skew-x and skew-y to
>   consolidated transform spec [recorded in
>   [22]http://www.w3.org/2010/03/11-fx-minutes.html#action04]
>
>   <trackbot> Sorry, couldn't find user - Simon
>
>   ACTION; Doug set up an FX wiki
>
>   <shepazu> ACTION: smfr to remove skew and add skew-x and skew-y to
>   consolidated transform spec [recorded in
>   [23]http://www.w3.org/2010/03/11-fx-minutes.html#action05]
>
>   <trackbot> Sorry, couldn't find user - smfr
>
>   trackbot, self.reboot()
>
>   <trackbot> Sorry, ChrisL, I don't understand 'trackbot,
>   self.reboot()'. Please refer to
>   [24]http://www.w3.org/2005/06/tracker/irc for help
>
>     [24] http://www.w3.org/2005/06/tracker/irc
>
>   <smfr> trackbot, reload
>
>   <dino> i thought there was a reload command
>
>   Resolution: Freeze the separate CSS and SVG specs, copy to new FX
>   repository
>
>   adjourned
>
>   <shepazu> ChrisL: couldn't hear you
>
>   <shepazu> try again?
>
>   <anthony> Issue: Need to figure out how transitions affect a
>   transform on an element that has an animation running on it
>
>   <trackbot> Created ISSUE-2 - Need to figure out how transitions
>   affect a transform on an element that has an animation running on it
>   ; please complete additional details at
>   [25]http://www.w3.org/Graphics/fx/track/issues/2/edit .
>
>     [25] http://www.w3.org/Graphics/fx/track/issues/2/edit
>
>
>
> Summary of Action Items
>
>   [NEW] ACTION: Chris to Simon to remove skew and add skew-x and
>   skew-y to consolidated transform spec [recorded in
>   [26] http://www.w3.org/2010/03/11-fx-minutes.html#action03 ]
>   [NEW] ACTION: Doug to create an FX repository [recorded in
>   [27] http://www.w3.org/2010/03/11-fx-minutes.html#action01 ]
>   [NEW] ACTION: Simon to remove skew and add skew-x and skew-y to
>   consolidated transform spec [recorded in
>   [28] http://www.w3.org/2010/03/11-fx-minutes.html#action02 ]
>   [NEW] ACTION: Simon to remove skew and add skew-x and skew-y to
>   consolidated transform spec [recorded in
>   [29] http://www.w3.org/2010/03/11-fx-minutes.html#action04 ]
>   [NEW] ACTION: smfr to remove skew and add skew-x and skew-y to
>   consolidated transform spec [recorded in
>   [30] http://www.w3.org/2010/03/11-fx-minutes.html#action05 ]
>
>   [End of minutes]
>
>
> --
>  Chris Lilley                    mailto:chris@w3.org
>  Technical Director, Interaction Domain
>  W3C Graphics Activity Lead
>  Co-Chair, W3C Hypertext CG

I hate skew too!

Now that a change wouldn't cause a split between the SVG and CSS
drafts, it would be great to take a second look at replacing skew()
with shear(), which is more idiomatic and a bit more sane wrt
animation without that hidden tan function. DBaron originally brought
it up last December:

http://lists.w3.org/Archives/Public/www-style/2009Dec/0059.html

Received on Thursday, 11 March 2010 22:46:37 UTC