Intent to unship: mozmm CSS unit.

[Emilio Cobos Álvarez]

Hi,

In bug 1416564 I intend to remove the mozmm CSS unit.

This unit is Mozilla-only, has no spec, and is unused in all our
codebase (except for two tests, one of those which tests the unit itself).

This unit was introduced experimentally in bug 537890, our browser
chrome code used it in bug 588464, and all that is gone since then.

Given there's no spec, and no usage in the wild as far as I can tell, I
think we should try to remove it.

Thoughts?

-- Emilio

[Xidorn Quan]

IIRC, we have discussed unshipping this somewhere before we added its
support to stylo (maybe the Taipei meeting this year?) and dbaron said
that mozmm provides an ability to size something based on physical size
which is not directly possible in any other unit, so this is something
we may want to put into spec rather than removing.

I doubt if there is anything changed since then (except that we still
haven't pushed this on csswg), so maybe we still shouldn't remove it.

- Xidorn

[Emilio Cobos Álvarez]

On 11/12/2017 05:14 PM, Xidorn Quan wrote:
> IIRC, we have discussed unshipping this somewhere before we added its
> support to stylo (maybe the Taipei meeting this year?) and dbaron said
> that mozmm provides an ability to size something based on physical size
> which is not directly possible in any other unit, so this is something
> we may want to put into spec rather than removing.

Huh, I don't recall being part of that discussion, but that sounds
plausible.

> I doubt if there is anything changed since then (except that we still
> haven't pushed this on csswg), so maybe we still shouldn't remove it.

I'm definitely ok with adding something like that if there's an use-case
and demand for it... But given I haven't been able to find any actual
usage of it, not even internal, I suspect there isn't much?

In any case, having features exposed to the web without any
specification or other browsers supporting it looks like the wrong thing
to me, and the kind of thing that may bite in the future (and that makes
Servo's life harder, too), so I still would like to unship it if possible.

If afterwards we figure out that this is needed, I'd be happy to see it
added as something every browser agrees with. My patch keeps the ability
to convert physical millimeters to CSS pixels, so it'd be trivial to
reintroduce if Mozilla decides to push for this feature and other
browsers agree.

-- Emilio

[Gijs Kruitbosch]

On 12/11/2017 16:36, Emilio Cobos Álvarez wrote:
> On 11/12/2017 05:14 PM, Xidorn Quan wrote:
>> I doubt if there is anything changed since then (except that we still
>> haven't pushed this on csswg), so maybe we still shouldn't remove it.
>
> I'm definitely ok with adding something like that if there's an use-case
> and demand for it... But given I haven't been able to find any actual
> usage of it, not even internal, I suspect there isn't much?

Related, but not exactly the same (so please carry on with whatever you
want to do with mozmm): browser chrome at least has a usecase for a CSS
unit that matches 1 device pixel on various dpis, because layout already
adjusts some (but not all!) sizes to align to device pixels and this
causes issues when we basically want "make a line that is 1 device pixel
wide". We can workaround the lack of such a unit with CSS variables and
something like:

@media (min-resolution: 2dppx) {
--mydevicepixel: 0.5px;
}

but it's tedious to do for dpi on windows (which has a lot more
variations than on macOS) and probably a perf nightmare.

~ Gijs

[Gijs Kruitbosch]

On 12/11/2017 16:56, Gijs Kruitbosch wrote:
> because layout already adjusts some (but not all!) sizes to align to device pixels

For reference: https://bugzilla.mozilla.org/show_bug.cgi?id=477157 and
friends.

~ Gijs

[L. David Baron]

On Sunday 2017-11-12 16:11 +0100, Emilio Cobos Álvarez wrote:
> Hi,
>
> In bug 1416564 I intend to remove the mozmm CSS unit.
>
> This unit is Mozilla-only, has no spec, and is unused in all our
> codebase (except for two tests, one of those which tests the unit itself).
>
> This unit was introduced experimentally in bug 537890, our browser
> chrome code used it in bug 588464, and all that is gone since then.
>
> Given there's no spec, and no usage in the wild as far as I can tell, I
> think we should try to remove it.
>
> Thoughts?

So I think this unit had a pretty strong use case: designing of
touch user interfaces, where touch targets need to be at least a
certain physical size in order to work well with human fingertips.

There's also a risk from having physical units in CSS that we
learned the last time CSS had them: designs that work with some
ratios of physical units to other units and break with other ratios.

I've been meaning to dig up the minutes from the time we tried to
get the CSS working group to add these units, but I haven't had a
chance over the last few days. I think if those minutes suggest
that the working group was receptive to adding such units to the
spec (or perhaps even agreed to do so, but then they were never
actually added by the editor), I'd tend to think we should leave our
implementation, whereas if the minutes suggest that the balance of
working group opinion was against them, then we should remove it.
Would you be able to try searching for those minutes? I know
there's been an in-person discussion (and I think it was at a
meeting at a TPAC a few yeas ago).

-David

--
𝄞 L. David Baron http://dbaron.org/ 𝄂
𝄢 Mozilla https://www.mozilla.org/ 𝄂
Before I built a wall I'd ask to know
What I was walling in or walling out,
And to whom I was like to give offense.
- Robert Frost, Mending Wall (1914)

[Emilio Cobos Álvarez]

Hi David,

So, I've looked through and I haven't found any related minutes (I've
looked for minutes that mentioned "physical", of which they were many,
but all related to physical properties and scroll snapping, and for
minutes that mentioned mozmm, of which they were none).

I've found two threads in the w3c mailing list that look related, one
which is basically the discussion that lead to all units being changed
to be relative to the density of the display:

https://lists.w3.org/Archives/Public/www-style/2010Jan/0058.html

And someone interjecting on a thread about adding the Q unit:

https://lists.w3.org/Archives/Public/www-style/2013Nov/0302.html

Then there's a relevant, more recent, csswg-drafts issue:

https://github.com/w3c/csswg-drafts/issues/614

But from the comments in there it doesn't seem to be a strong sense
about it being useful, but mostly the opposite (and actually [1] may be
a legit bug we should look into if we keep it in tree, or behavior we
should document). Also, some other alternatives for that use case that
work in all browsers are mentioned, like resolution-dependent media queries.

What do you think David? I landed the patch this morning, and I'd really
prefer it to stick. Given the recent cssswg-drafts issue where the CSSWG
members were not really excited about it, I think it's ok to remove it,
but happy to back the patch out if you think it's not.

-- Emilio

[1]: https://github.com/w3c/csswg-drafts/issues/614#issuecomment-254679777

>
> -David
>

[Jonathan Kew]

On 15/11/2017 03:32, Emilio Cobos Álvarez wrote:

> So, I've looked through and I haven't found any related minutes (I've
> looked for minutes that mentioned "physical", of which they were many,
> but all related to physical properties and scroll snapping, and for
> minutes that mentioned mozmm, of which they were none).
>
> I've found two threads in the w3c mailing list that look related, one
> which is basically the discussion that lead to all units being changed
> to be relative to the density of the display:
>
> https://lists.w3.org/Archives/Public/www-style/2010Jan/0058.html
>
> And someone interjecting on a thread about adding the Q unit:
>
> https://lists.w3.org/Archives/Public/www-style/2013Nov/0302.html

It sounds like you may have missed the lengthy "[css3-values] Physical
length units" thread (spun off from "[css3-mediaqueries] DPI in
resolution media queries") that started in mid-Feb 2012:

https://lists.w3.org/Archives/Public/www-style/2012Feb/0627.html

I haven't re-read all that thread to see whether anything resembling a
consensus seemed to be emerging.... my (vague) recollection is that
there's generally little enthusiasm in the WG for a "true" physical
unit, with perhaps one or two vocal dissenters.

JK

[Emilio Cobos Álvarez]

On 11/15/2017 11:51 AM, Jonathan Kew wrote:
> On 15/11/2017 03:32, Emilio Cobos Álvarez wrote:
>
>> So, I've looked through and I haven't found any related minutes (I've
>> looked for minutes that mentioned "physical", of which they were many,
>> but all related to physical properties and scroll snapping, and for
>> minutes that mentioned mozmm, of which they were none).
>>
>> I've found two threads in the w3c mailing list that look related, one
>> which is basically the discussion that lead to all units being changed
>> to be relative to the density of the display:
>>
>>    https://lists.w3.org/Archives/Public/www-style/2010Jan/0058.html
>>
>> And someone interjecting on a thread about adding the Q unit:
>>
>>    https://lists.w3.org/Archives/Public/www-style/2013Nov/0302.html
>
> It sounds like you may have missed the lengthy "[css3-values] Physical
> length units" thread (spun off from "[css3-mediaqueries] DPI in
> resolution media queries") that started in mid-Feb 2012:

Indeed, thanks Jonathan! I bet the minutes at

https://lists.w3.org/Archives/Public/www-style/2012Feb/0530.html

were what David was referring to.

[Robert O'Callahan]

I thought there was also a legitimate use-case for displaying content "life
size", e.g. if you wanted to draw a ruler on a tablet.

But if the CSSWG doesn't agree after all this time, just drop it I guess.

(Though I think there's something slightly broken about how Web developer
needs are bubbling up to WGs. For example GeometryUtils hasn't been
implemented by other browsers, who apparently detect no pressure from Web
developers to solve the use-cases it solves, e.g. computing px offsets
between arbitrary elements even if they're in a DOM subtree with a scale
transform. Yet I ran into that problem pretty quickly while coding a Web
UI. Maybe I'm just strange...)

Rob
--
lbir ye,ea yer.tnietoehr rdn rdsme,anea lurpr edna e hnysnenh hhe uresyf
toD
selthor stor edna siewaoeodm or v sstvr esBa kbvted,t
rdsme,aoreseoouoto
o l euetiuruewFa kbn e hnystoivateweh uresyf tulsa rehr rdm or rnea
lurpr
.a war hsrer holsa rodvted,t nenh hneireseoouot.tniesiewaoeivatewt sstvr
esn