See also: IRC log
<trackbot> Date: 27 March 2014
<scribe> scribenick: cabanier
ed: I was looking at a bug for
text rendering in blink
... for some fonts, the bouding box was not encompassing the
text
... the filter that I was applying, didn't cover the text. I
found that the spec is unclear
... I was hoping to get agreement and clarification on glyph
cells and bounding boxes on text elements
... if you have a font where the glyphs go outside the advance
width
<ed> https://www.google.com/fonts/specimen/Bangers
ed: some browser use this for the bounding box
heycam: I've used glyphs that go outside the box
ed: if you look that page and
select the text, you will see that the last charactor is not
selected
... my first question is: the x position of the glyph cell,
should that always be 0 or can it be a negative value?
heycam: I can't remember the term
in the font
... weren't we going to take the glyph cells and union that
with the ink boundary?
tav: this is quite complicated
because kerning can move the characters
... the box is a good starting point, but maybe you need an
additional margin
ed: I would like to see a boundingbox that tightly encompasses the glyphs shapes
tav: a font designer spends a lot
of time thinking about htis
... I'm not sure that using the painted region is what you
want
... because the text shouldn't always change the dimension of
the box
... give what the font designers give you and then add
padding
ed: for first last characters? all the glyphs?
tav: the height is always the
same
... this makes it an easy calculation
cabanier: doesn't the font have the bounding box?
Tav: that is not always the
correct one
... I'm not sure if you can do better by looking at the painted
region
heycam: maybe you want to make sure that hit testing uses the painted area of the text
ed: what would you base the padding on?
heycam: the author would set it
ed: that doesn't work for web fonts since you don't know if you get it or if there's a fallback one
Tav: it should be pretty simila
heycam: if there's overflow etc. maybe the padding should be based on the worst character of the font
Tav: if you have a stem going down the line, you don't want them to be different height
ed: when it comes to the width, there's a difference between the browser
<ed> <text filter="url(#obb-filter)">some slanted text</text>
ed: obb-filter could put the text back, you don't get all the text back
cabanier: and this is a spec bug?
ed: it's not clear enough
... we never define the term of glyph cell
cabanier: advance is clearly incorrect
ed: you could look at things
heycam: you will have to look at
the glyphs in the glyphs in case they overflow a lot too
... this does need extra calculations
ed: some browsers do it the way I
expect
... so they have something like a painted bounding box
heycam: we could add an extra flag to opt into that
cabanier: not as the default
heycam: would we still union it so it's possible to make it bigger?
<Tav> http://fontforge.org/overview.html#Baseline
heycam: would you like to have it controlable?
ed: that sounds like a good thing. I would like the default to cover the rendered area
heycam: do you know what the different browsers do?
ed: I have a test case. FF,
presto gave a bbox that encompassed the rendered glyphs fully.
Chrome did not.
... (looking for test case)
<heycam> "right"
:-)
ed: Firefox and presto had the
full rendered bounding box of the text run
... I'll bring the text case to the F2F
heycam: I can see 3 different
behaviors
... 1 = exact glyph shapes
... 2 = union of glyph cells and the bounding box of each
glyph
... 3 = just the glyph cells
... hit testing would be interesting as well
ed: a 4th one would include text decorations
cabanier: those would be needed for filters as well?
Tav: a 5th one would include padding
ed: I don't think we define
anything for how hit testing on text is done
... it's up to the ua to figure this out
heycam: I think hit testing for mouseover was the reason we added the glyph cell approach
ed: text rendering, should that
affect the result?
... you might get different bounds for different rendering
modes
... we don't call this out in the spec
heycam: it probably should
ed: these are all the questions I had in the mail
heycam: what should be the default and how many options should be offer?
ed: yes, and define the correct term
heycam: let's make a decision at the F2F
richardschwerdtfeger: we had a
meeting yesterday on what spec type to create (respec,
etc)
... whose IP policy do we follow?
... where do publish it? PF or SVG?
... what does the group think about that
heycam: Doug would be the best
one to know
... how does the FX group work?
ed: the IP policy is for both of the groups
richardschwerdtfeger: it could be
a joint publication
... I do know that HTML5 implementation
... there's a core ARIA spec that says how to map the a11y
API
... and then there's a HTML 5 spec that references that
... each platform implements it differently
heycam: my feeling is that the PF
people will do most of the work
... so it feels like it makes most sense to do it there
richardschwerdtfeger: does that mean that members need to join in order to contribute?
heycam: I'm unsure
richardschwerdtfeger: let me check with Doug
heycam: Chris is the official
team contact
... I would like to know from them what is the best
approach
richardschwerdtfeger: you're OK if it's done by PF
heycam: let me look into that
richardschwerdtfeger: the group agrees that the PF group does the work. I will ask if people would have to join the PF group if they want to contribute
http://lists.w3.org/Archives/Public/www-svg/2014Mar/0008.html
heycam: this was a request from
the mailing list
... the proposal was for an additional processing attribute
called 'media'
... the first one that matches would be rendered
... is this a good idea? If so, what timeframe?
<heycam> https://www.w3.org/Graphics/SVG/WG/wiki/Proposals/IFrame_Like_Syntax#5.9.2_The_.27switch.27_element
heycam: I don't remember the
decision
... my feeling is that it's a simple thing to add
... and it would be Ok to add in the SVG 2 timeframe
cabanier: you can't just do this by using CSS?
heycam: it's slightly tricky. you would have to combinate them somehow
cabanier: it feels like CSS should solve that
heycam: maybe
... you could imagine that something like that exists
... but that would be similar to adding it to the markup
... it feels like a natural place to put it
... let's discuss it at the F2F
ed: doesn't sound like a bad idea but I'm not a fan of the switch element
heycam: in HTML some elements
have the media attribute
... in SVG, it seems like a reasonable parallel
... let's keep it on the agenda of the F2F
... next week we won't have a meeting
This is scribe.perl Revision: 1.138 of Date: 2013-04-25 13:59:11 Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/ Guessing input format: RRSAgent_Text_Format (score 1.00) Succeeded: s/encompasses/tightly encompasses/ Succeeded: s/correctly/the way I expect/ Succeeded: s/did it right/gave a bbox that encompassed the rendered glyphs fully/ Found ScribeNick: cabanier Inferring Scribes: cabanier Default Present: [IPcaller], ed, heycam, cabanier, stakagi, Tav, Rich_Schwerdtfeger, birtles Present: [IPcaller] ed heycam cabanier stakagi Tav Rich_Schwerdtfeger birtles Regrets: Brian Dirk Agenda: http://lists.w3.org/Archives/Public/public-svg-wg/2014JanMar/0139.html Found Date: 27 Mar 2014 Guessing minutes URL: http://www.w3.org/2014/03/27-svg-minutes.html People with action items:[End of scribe.perl diagnostic output]