SVG Working Group Teleconference -- 02 Jun 2016

<stakagi> hello

<scribe> scribe: Nikos

<scribe> scribenick: nikos

Switch and conditional processing issues

AmeliaBR: There's one issue as a summary, with lots of specific issues hanging off that

<AmeliaBR> Summary: https://github.com/w3c/svgwg/issues/139

switch originally had three different attributes for controlling when you switch different features

scribe: one is requiredExtensions, think the idea was that if somebody somewhere in some other spec or software defined an extension, then in svg we could define something that should only be rendered if that was supported
... it doesn't seem to be well used
... but some things would be useful to define
... like whether this svg ua can render html
... have a non standard extension in FireFox that takes the xml namespace
... not defined in the standard anywhere, and not supported by other browsers
... think it would be useful to have some way of saying what type of documents this UA supports
... or suggest what type of image file can you embed?

nikos: I think I'd like to defer this.
... can we put this in the integration spec?
... or another spec that's just a list of supported extensions?
... I'm supportive of capturing the extensions that are used by user agents so we can get interop

shepazu: I thought integration spec too

AmeliaBR: SVG 2 just defines the content as a url and doesn't have requirements beyond that

RESOLUTION: SVG WG will list details of common extensions and supported values for the requiredExtensions attribute in the integration spec

nikos: Would it be better to just support the picture element rather than image formats in switch?

AmeliaBR: That's not as backwards compatible as switch
... right now, you could get an image if you used switch so you would get a sensible format

shepazu: Is switch supported in illustrator or Inkscape?

AmeliaBR: it's supported in rendering afaik but you can't generate it

Tav: I'm not 100% sure, I know we parse it but not sure what we do with it
... think there are comments saying we should do something or other with it

shepazu: When I think about SVG, there's a number of places, I think of authoring tools and browsers primarily
... there's other environments, but I don't expect they're using switch at all
... so when you say it's not supported, I think it would be trivial to do this in an svg + html context
... and since modern browsers support html5, I wonder what part of the ecosystem wouldn't support picture?

AmeliaBR: it's not supported right now for svg

Tav: not supported by Inkscape or Illustrator

shepazu: I think if we are going to change behaviour, we should aim towards html rather than going alone

AmeliaBR: it's also relevant to foreignObject, to define what time of content should be supported

shepazu: my position is generally to not add new features or not extend less well supported features of SVG

nikos: Agree

systemLanguage attribute

<AmeliaBR> https://github.com/w3c/svgwg/issues/136

AmeliaBR: There's a bit of a limitation in how systemLanguage is defined
... assumes users only have one preferred language
... or that all options are equal
... this can be problematic if users have for example preferred languages, but will accept English if need be
... SMIL in later versions introduced an attribute that allows the user agent to be smarter
... hasn't been adopted into svg
... is supported in FireFox
... though there's HTML parsing issues because it's a mixed case attribute
... my pref is to spec so smart language matching is the default
... and not worry about the attribute, because I don't see when you'd want the other behaviour

nikos: I support this. Language choice is one of the more useful uses of switch
... it's defined already in another spec so we can just reference it
... are there any issues changing the default behaviour?

AmeliaBR: are people leaning towards adding the reoder attribute? or define language matching?

Tav: the latter

AmeliaBR: In FireFox you have to specifically allow reorder, the default is the old behaviour

shepazu: as long as we can get implementers, I like the second option

RESOLUTION: Language selection order for systemLanguage attribute will follow allowReorder=true by default

<scribe> ACTION: shepazu to have a go at changing systemLanguage to allowReorder by default [recorded in http://www.w3.org/2016/06/02-svg-minutes.html#action01]

<trackbot> Created ACTION-3844 - Have a go at changing systemlanguage to allowreorder by default [on Doug Schepers - due 2016-06-09].

shepazu: would be best to include inline rather than reference SMIL

no title/desc in switch

AmeliaBR: think this is just editorial
... title and desc interfere with the selection of child elements so we resolved to remove them from the content model

Switch with non rendered (audio and visual ) content

UNKNOWN_SPEAKER: There was a previous resolution that switch doesn't affect script and style
... we had an issue raised about how it interacts with audio and video
... my suggestion was that it should be treated as if audio and video are display:none
... which means the audio and video don't play

nikos: I agree with that behaviour. Was trying to find where in HTML5 it says audio and video don't play on display none
... if anyone knows where that is, drop a link in the issue please
... There was the question of whether resources should be loaded at all. But they probably have to, because they can be accessed through the DOM. It's like images that are display:none

removal of requiredFeatures

In SVG 1.1 we had requiredFeatures that had a defined list or category of features that svg would support

scribe: that has been removed from SVG 2
... browsers implement it in a way where they didn't behave properly if you put in something that wasn't recognised
... I was wondering whether it might make sense to add back in as a deprecated feature
... so UA must support all SVG 11 strings

nikos: was wondering if we should leave it in and define it as returning true

AmeliaBR: that's the same as removing because of the behaviour defined in the spec.

nikos: that's ok then, probably the reason why it was just removed. I say just leave it as is

shepazu: I agree with that
... browser vendors in the past have always pushed back on conditional processing. Unless that has changed in the last couple of years, I don't see us getting momentum in the implementation so don't think we should put much effort in

AmeliaBR: CSS conditional processing seems to be doing well, I'd love to build on that but not something we can do now

shepazu: I tried to put something like this in dom3 events but was met with strong resistance
... they're not bad ideas, just need to focus on what we need to do to get SVG 2 done

AmeliaBR: agree with new features, what do we do with existing features that are poorly implemented? How do we clean up the spec?
... I have lots of ideas how to make switch cool with MQ and stuff

shepazu: If a feature was poorly implemented, it could be because it was poorly specced, it could be that it wasn't tested well, could be lack of browser interest
... we need to work out on a case by case basis which one it was
... for switch, it never really passed the critical mass of implementers that it needed to succeed
... understand what you're saying about what do we do with what's left in the spec

AmeliaBR: switch and requiredFeatures has two implementations unless it's been torn out in the last year
... problem is the feature strings aren't clearly defined
... testing for filter support for example didn't mean the feature was supported without bugs

shepazu: that's why hasFeature fell by the wayside
... the fine grainedness was missing
... browsers don't want to implement that level of grain
... I say this as a person who strongly advocated for this in the past

AmeliaBR: it might be useful to have a note in the conditional processing overview saying the requiredFeatures attribute has been removed

<scribe> ACTION: Amelia to add a note regarding removal of requiredFeatures [recorded in http://www.w3.org/2016/06/02-svg-minutes.html#action02]

<trackbot> Created ACTION-3845 - Add a note regarding removal of requiredfeatures [on Amelia Bellamy-Royds - due 2016-06-09].

- DRAFT -

SVG Working Group Teleconference

02 Jun 2016

Attendees

Contents

Switch and conditional processing issues

systemLanguage attribute

no title/desc in switch

Switch with non rendered (audio and visual ) content

removal of requiredFeatures

Summary of Action Items

Summary of Resolutions

Scribe.perl diagnostic output