igari: problems
... can't validate content modified by JS
... Can't validate computed styles
... can't validate matching of style and style for each HTML
Tag
... while validation should be done automatically
... as you know, there is the W3C CSS Validation Service
... @@@v
igari: also Jet Brains
... case study
... case 1-1: needless styles
... Q: value of display property of element?
... that is block
... unnecessary style should be removed
... but there is some potential harm with removing it
... case 1-2: needless styles
... Q: What is value of display property of element?
... @@@p
daniel: value of display property of
is block?
... gecko returns inline
... chrome returns block
... maybe it's a bug of chrome
igari: yes, I used chrome to check
this...
... moving ahead
... case 1-3: needless styles
... Q: What is value of 'width' property of element?
... A: auto
... Case 2-1: Risky Styles
... Q: Pseudo element into empty element?
... A: this is buggy
... case 2-2: Risky Styles
... case 3-1: Mistake in Media Query
... Q: no parent table-cell in media query
... case 3-2: mistake after JS
... Q: after inserting DOM by JS
... invalid styles give you: meaningless patch, interrupting
creative ideas,...
... resolution!
<scribe> ... new style validator
UNKNOWN_SPEAKER: validation on the browser
(demo)
igari:
stylev/Style-Validator/page/demo-simple.html
... Style Validator can detect invalid style
... some JS script is installed and we can validate the CSS
dynamically
... Introduction
... this is a chrome extension
... but this mechanism has some week points
... not an official standard
... no document available and depends on each developers who use
this
... Validation Rule Site
... can manage the rules for validation here
... and save the configured rules as a JSON format
daniel: question on values within selectors
(write down an example on the white board)
[[
.foo {
content; attr(p.bla, title);
}
]]
daniel: what if there is no p.bla defined?
igari: can't detect that
currently
... but would try to implement if there is a need
takeshi: does this tool have a capability of reporting?
igari: no at the moment
... but am planning to add reporting capability
myles: what kind of mechanism you use to get the style?
igari: using getComputedStyle and
getUncomputedStyle
... the latter is my own developed function
... getUncomputedStyle uses getComputedStyle and some CSS rules
myles: daniel gave an example
... and wondering about the future capability
... intend to check everything
igari: would feedback from you :)
daniel: this kind of style check is
not syntax check
... so it's a manual research and a bit difficult
... can all the requirements expressed in a machine-readable
manner?
igari: would make this tool an open source and get help from people
daniel: what is your motivation/rationale for this work?
igari: this is based on real use cases coming from my company's colleagues
adachihara: wondering about the possible validator for the validation rule tool
igari: yes, the rule tool has that capability
adachihara: wondering about multiple the case with multiple cases match
@1: is the target of this tool a developer, or somebody doesn't have CSS knowledge?
igari: currently developers
@1: publishers, etc., who don't have CSS knowledges also may want to use this
ohata: who manages the rules?
<myles> glazou: is there a way to get specified style in CSS? i don't know
igari: there are already pre-defined
rules within the validation rules tool
... so you don't have to specify the rules
... if you need to add some tweaks, you can
ohata: is this already publicly available?
<glazou> myles: you mean from a given element ?
<myles> yes
<myles> glazou: glazou: i guess there would be many specified styles depending on selector specificity
igari: not yet. but planning to publish by the end of this year
jeff: does this tool parse the CSS?
<glazou> short answer is no ; there used to be private interfaces for that in Gecko but they were removed ; you could always of course work from the OM of the stylesheets, find the rules matching and do the first part of the cascade by yourself
igari: no, but this tool use the browser's parsing results
<glazou> expensive but doable
<myles> glazou: you would have to evaluate the selectors yourself
<myles> (if you went with the CSSOM approach)
<glazou> only thing you have to do is call matchSelector() :-)
<myles> yes, yes :)
<glazou> BUT you have to parse the selector to compute specificity since, damn it, we never provided an API for that
igari: and check the validness based on that
<glazou> and then balance all specificities and importance for rules that apply
<glazou> and get the result of that
daniel: if somebody implements a browser-based inspector inside (=natively) browser, what do you think?
igari: can't answer at the moment...
daniel: wondering users would be
happy with this tool
... there is legal instructions for accessibility
... but no restriction for CSS validation
igari: that's true
kaz: maybe even more useful if authoring tools have this capability
[ adjourned ]