Spec defines 'double' in a way that collides with WebIDL

[hober]

Maybe use a different term?

I'm seeing this Bikeshed error on an unrelated document:

LINK ERROR: Multiple possible 'double' idl refs.
Arbitrarily chose https://w3c.github.io/edit-context/#dom-underlinestyle-double
To auto-select one of the following refs, insert one of these lines into a <pre class=link-defaults> block:
spec:edit-context; type:enum-value; text:double
spec:webidl; type:interface; text:double
{{double}}

[dandclark]

@snianu or @alexkeng do you know where "double" would actually be needed? In Chromium at least it doesn't look like "double" is a possible IME underline style (link):

  enum class UnderlineStyle {
    kNone,
    kSolid,
    kDot,
    kDash,
    kSquiggle,
  };

I'm wondering if this can just be removed...but maybe you remember some context for why it's an option in the spec? Is it used with IMEs in other browsers?

[dandclark]

I spoke with @snianu about this today. double was added to EditContext early on because it was expected to be used by a planned Windows feature, but that feature never landed. So at least for now there is no known use for it. I will Agenda+ to discuss next WG meeting but I expect that this can simply be removed.

[css-meeting-bot]

The Web Editing Working Group just discussed Spec defines 'double' in a way that collides with WebIDL #89.

The full IRC log of that discussion

<dandclark> topic: Spec defines 'double' in a way that collides with WebIDL #89
<dandclark> github: https://github.com//issues/89
<smaug> dandclark: double isn't actually hooked up with anything, so there is no known use for it
<smaug> dandclark: if it is not used, we should just remove it
<smaug> johanneswilm: is use cases appear in the future, we can add more types
<smaug> johanneswilm: anyone fine removing it
<smaug> yes
<smaug> https://github.com//issues/88
<smaug> topic https://github.com//issues/88
<smaug> dandclark: the spec doesn't handle the bounds of the selection start/end
<smaug> dandclark: should we clamp or throw ?
<smaug> smaug: Range does throw
<smaug> johanneswilm: throwing error could make it easier to find issues in js code
<smaug> johanneswilm: it would be good to give a way to find js issues
<smaug> smaug: are there APIs which do clamp?
<smaug> dandclark: maybe the right answer is to throw
<smaug> dandclark: hopefully we can find that no one relies on the current behavior
<smaug> dandclark: and also if we can find counter example where clamping does happen
<smaug> smaug: by default we should try to follow existing old APIs
<smaug> dandclark: I'll investigate