SVG Working Group Teleconference -- 27 Apr 2009

<trackbot> Date: 27 April 2009

<shepazu> nope

<scribe> scribe: erik

<scribe> scribeNick: ed

Publishing the Compositing spec

CM: we discussed previously to publish the spec this week, sent some comments today, but those doesn't need to hold up the publication

AG: those comments are good, mostly typo corrections, and a couple of things that need to be explained
... as long as everyone else is ok with publishing

DS: has it been FPWD?

AG: this would be the FPWD

CM: parts were published before in the 1.2 full draft

DS: this is a new spec though

CM/DS: we can do the cleanups/typos after the FPWD

CM: are we resolved to publish compositing?

DS: any objections?

all: no objections, let's publish

RESOLUTION: to publish SVG Compositing as a FPWD

<scribe> ACTION: DS to publish the SVG Compositing spec [recorded in http://www.w3.org/2009/04/27-svg-minutes.html#action01]

<trackbot> Created ACTION-2527 - Publish the SVG Compositing spec [on Doug Schepers - due 2009-05-04].

Updating broken CSS2 links

DS: doesn't need discussion?

CM: we need to decide how to deal with it, in the existing spec
...summary: because of the shortname /CSS2 and /REC-CSS2 are now redirected to CSS21, various links in SVG 1.1 are now dangling
... full dated uri:s should have been used from the start in the spec
... we can replace the SVG 1.1 spec links in place to fix the links

ED: are we still pointing to CSS2, or 2.1?

CM: css2
... if there are no objections, we can fix the links, have link to the files that need to be switched
... there are a couple in tiny 1.2 too, not broken ones, but it's now pointing to css21...and those features are still in css21 anyway
... but css2 is in the ref-section
... will forward the mail with the fixups

<scribe> ACTION: heycam to mail DS the info for fixing up CSS2 links in the SVG 1.1 spec [recorded in http://www.w3.org/2009/04/27-svg-minutes.html#action02]

<trackbot> Created ACTION-2528 - Mail DS the info for fixing up CSS2 links in the SVG 1.1 spec [on Cameron McCormack - due 2009-05-04].

Referencing CSS 2.1 from SVG 1.1 Second Edition

CM: I'm not that familiar with what's changed
... one thing I know of is the webfonts stuff, which is now in CSS3 webfonts

DS: I don't think we should change SVG 1.1 that much

CM: probably more effort than its worth
... we don't want to change things, but rather to point to the more up-to-date spec (CSS21)
... they may be in CR for awhile anyway

DS: it wouldn't be correct to change it
... we may be changing things dramatically for older svg useragents
... we should definately reference CSS21 going forward, but not in the SVG 1.1 second edition

ED: yes, I agree with that

CM: fine with me

Pattern and rendering order

<heycam> http://lists.w3.org/Archives/Public/www-svg/2009Apr/0088.html

<heycam> ED: he's asking about the case when the pattern tile has some overflow, asking how it's supposed to render

<heycam> ED: of course, UAs differ on how they interpret this

<heycam> ED: i think he's saying that the most logical solution would be to use the painters model

<heycam> ED: he's saying that because at least opera and asv don't support overflow, you can't do all types of overlapping patterns

http://lists.w3.org/Archives/Public/www-svg/2009Apr/0093.html

<heycam> ED: he refers to the 17 known tiling patterns

<heycam> ED: in my response i ask him what kinds of patterns couldn't you do

<heycam> ED: he replied with those 17 groups

<heycam> ED: because you can do quite a few types by rearranging the tile itself

<heycam> ED: if you want to be able to do all these other ones, then the pattern rendering would have to be a bit more advanced

<heycam> ED: the 1.1 spec doesn't really say how you render those cases, it just says that you can have overflow:visible, doesn't say what that's supposed to look like or handled

<heycam> ED: i haven't tested that many implementations, it would probably be nice to have examples of each of these 17 groups of periodic tilings, and to see what all of the current UAs do

<heycam> ED: one problem would be that there are not that many UAs that support patterns

<heycam> ED: i don't think safari does it

<heycam> ED: batik, asv, opera and i think firefox 3

<heycam> ED: those should be tested for all those groups

<heycam> ED: i think batik/asv/opera all just take the rectangle and tile that

<heycam> ED: i think for 1.1 Second Edition we should disallow it, or state that it's undefined explicitly

<heycam> CM: would we say that overflow has no effect, and that it always clips to the pattern viewport?

<heycam> ED: or we could say that it's undefined behaviour, so we could leave it open for fixing later

<heycam> CM: is it worth having it undefined so that we can define it later?

<heycam> DS: it's worth considering

<heycam> AG: should email olaf back

<heycam> ED: we should test the fallback behaviour; it's possible it's just clipped to the viewport

<heycam> DS: this is definitely something we should consider following up on, not necessarily something we should change now

<heycam> DS: IME patterns are useless sometimes

<heycam> ED: so we could just say that overflow:visible is undefined

<heycam> AG: raise an issue on svg 2 to define it

<scribe> ACTION: ed to add an svg 1.1 errata for making overflow=visible on patterns be undefined behaviour and to raise an issue on solving the pattern tiling for overflow=visible (core2) [recorded in http://www.w3.org/2009/04/27-svg-minutes.html#action03]

<trackbot> Created ACTION-2529 - Add an svg 1.1 errata for making overflow=visible on patterns be undefined behaviour and to raise an issue on solving the pattern tiling for overflow=visible (core2) [on Erik Dahlström - due 2009-05-04].

�image-fit� vs preserveAspectRatio

ED: so the only remaining issue really is the naming it seems
... the feature is useful for both CSS and SVG

CM: so it would be fine if UAs implented both pAR and image-fit/content-fit

DS/CM/ED: we all like 'content-*'

<scribe> ACTION: ed to reply to the image-fit thread saying that the SVG WG would prefer the name(s) 'content-*' [recorded in http://www.w3.org/2009/04/27-svg-minutes.html#action04]

<trackbot> Created ACTION-2530 - Reply to the image-fit thread saying that the SVG WG would prefer the name(s) 'content-*' [on Erik Dahlström - due 2009-05-04].

ref/params spec

DS: any reviews so far, besides heycam?

AG: not yet

ED: only read the proposals so far

DS: heycam has proposed an alternate syntax
... a suggestion about an alternate syntax that got rid of the ref element, and directly put a functional values in the attribute values

CM: yeah, wanted to see if I could get rid of the ref elements

DS: yeah, we talked about it at the TPAC
... paintservers are similar
... would it be easier to implement if it used the same mechanism?
... thinking about it now, I could implement it in script either way

CM: the weird part in my proposal is the content attribute, which is similar to :content in css
... so it would be more general way of inserting text content

DS: one problem with tref, if the content doesn't resolve then the user is left unknowing, nothing will be shown (no fallback content)
... so I liked the idea of allowing fallback content for tref
... are we talking about markup or just strings?
... tref just takes a string
... having fallback for tref and having content, not so keen on having stirngs in attributes though

CM: in svg normally if you have a tref reference to a switch would it give you the textContent or would it do the switch processing?

ED: I think it's just textContent

CM: a problem is that one of the role attributes is named 'content'
... the reason I chose 'content' is that it offers the same functionality as the css content property
... what do you want in terms in of publication?
... might be good to publish this to let people know we want to add this functionality

AG: yeah, we should get it out and get comments

CM: DS you could add some notes there, for these things

DS: what do you think of changing the values of things like geometric values of things like x and y, to have functional values?

CM: you mean turning attributes into properties?

DS: did you see the primer?

<shepazu> http://dev.w3.org/SVG/modules/ref/SVGRef.html

<shepazu> http://dev.w3.org/SVG/modules/ref/SVGRefPrimer.html

DS: scroll down to the maps
... there's a little dot the, parameterising the cx and cy
... and same for links, to pass in links for e.g ads

CM: like the examples

DS: right now we cna only use functional notation for attribute values in presentation attributes
... i'd like to extend that to all attributes
... one, why wouldn't you want to set anything with css?
... two, it would work for camerons constraints stuff

CM: it will be important to think about how this would work for future things

DS: parameterised attribute values is probably how I would go with extending the syntax
... so I'd like to say e.g 100%-5px without having it in a URL

CM: calc in CSS provides that
... if we're looking at promoting the attributes to properties, then we could gain from that
... it makes enough sense to be able to put them there

<shepazu> http://www.w3.org/TR/css3-values/#calc

CM: it does make sense to upgrade them to being properties

ED: all attributes?

DS: most, href would be an exception

ED: lists of things e.g, sometimes you might want to override only parts of a list (or a path)

DS: anything that takes a list couldn't have a fallback
... any other attributes could have a fallback

CM: that might mean we have to think about types

DS: if we do it, it's done, there's going to be flaws with any model we come up with
... i like functional values
... it builds on things laid out in css

<shepazu> width: calc(100%/3 - 2*1em - 2*1px);

CM: i would be in favor of investigating making existing svg attributes into properties

DS: given that we want to allow this syntax anyway
... I think it makes sense for parameterising things like stroke and fill
... might make sense for other things (I think it does)
... less work to specialcase the ones that aren't properties

CM: would make it easier to reuse content

DS: markers is one thing, had to duplicate the markers several times to style them differently

ED: yeah, it's annoying that it behaves different from use, which do inherit style into the shadowtree

DS: i like having an ref element, because you can animate those in the document
... and we don't have the equivalent of <param> in svg
... if I wanted to use a button multiple times, I might like to actually like to use this syntax
... I was thinking of the <animation> element, or the <image> element, anything that references
... even the <use> element

AG: would we run into problems with optional arguments?

DS: <ref> have default values, but the example with the rect above, you could also pass defaults there
... it's important

AG: multiple optional elements
... or multiple optional arguments

DS: this is a new class of things we could start doing...calc, or url() and param
... there's an attr() function

CM: often used with css content

<anthony> transform="param(blah) translate(20 20)"

DS: we should talk to the csswg about this

AG: so, would the translate be an additonal transform, or a fallback?

<shepazu> transform="scale(param(blah)) translate(20 20)"

DS: it's a list, so would need to be specialcased

AG: needs special consideration yeah

DS: might be handy with translateX,Y,Z
... will add notes with the functional notation for param
... and if we're not going to have an element, then it could be applied to css as well

CM: yeah, that might be good

RESOLUTION: publish the param/ref spec as FPWD

<scribe> ACTION: heycam to get the ref spec building with the new scripts [recorded in http://www.w3.org/2009/04/27-svg-minutes.html#action05]

<trackbot> Created ACTION-2531 - Get the ref spec building with the new scripts [on Cameron McCormack - due 2009-05-04].

<scribe> ACTION: DS to integrate the new potential syntax, and publish the ref spec [recorded in http://www.w3.org/2009/04/27-svg-minutes.html#action06]

<trackbot> Created ACTION-2532 - Integrate the new potential syntax, and publish the ref spec [on Doug Schepers - due 2009-05-04].

<scribe> ACTION: heycam to look into making attributes into properties, and potential problems with that [recorded in http://www.w3.org/2009/04/27-svg-minutes.html#action07]

<trackbot> Created ACTION-2533 - Look into making attributes into properties, and potential problems with that [on Cameron McCormack - due 2009-05-04].

rrs-agent, make minutes

- DRAFT -

SVG Working Group Teleconference

27 Apr 2009

Attendees

Contents