Re: [Tools-discuss] [Cellar] non-ascii characters (fwd) Dave Rice: non-ascii characters

[Carsten Bormann <[e-mail]>]

On Aug 30, 2019, at 06:25, Julian Reschke <[e-mail]> wrote:
> 
> On 30.08.2019 06:20, Carsten Bormann wrote:
>> On Aug 27, 2019, at 23:06, Tom Pusateri <pusateri=[e-mail]> wrote:
>>> 
>>> RFC 7997 is pretty clear when UTF-8 is permissible and math symbols are not in the list.
>> 
>> I believe, with a 2019 perspective on the world, this is no longer the right decision.
> 
> Did the perspective change significantly within the last 4 years??

I think so.  Of course, tipping points are always a perception issue, but I think it would not be hard to obtain consensus today that we are past the UTF-8 tipping point.  Of course, there will be people who remember that they argued otherwise a decade ago and will try to maintain that position, but outside of vintage computer museums, the reality is pretty well-defined now.

>> We are incurring the costs of supporting Unicode but not reaping its benefits.
>> Providing HTML as a main viewing format and not even being able to solve the hyphen-dash-minus problem leads to weird-looking documents.
>> How is requiring SVG display support for understanding a document better than simply requiring Unicode support?
> 
> I do prefer proper hyphens, but I also note that the proper choice can
> lead to bikeshedding.

I’m not aware of an opportunity for that (well, maybe except for the choice between en-dashes and em-dashes for the punctuation mark in the form of a horizontal line used to indicate a pause, to delimit an inserted sentence or phrase, or to indicate a deliberately omitted word).    But given that this is the IETF, I’m sure we’ll find one :-)
More seriously, this is one place where copy-editing can simply solve the problem.

> My recollection is that we started with the use cases that seemed most
> important back then. I'll also note that Unicode math characters won't
> be sufficient in all cases, so MathML definitively is something to
> consider as well.

Having math characters available is not a valid replacement for something like MathML.
Being able to throw in some casual in-line math (that an authoring tool would convert to the right characters) is useful, though.  As in ⌈a⌉ and friends…  (Nobody would argue that I can’t use a*b in a spec to talk about the product of a and b, so why not a∉B or ∀x.x²≥0.)

Grüße, Carsten