--On Saturday, 27 February, 2021 23:48 +0100 Kjetil Torgrim
Homme <kjetilho@xxxxxxxxxx> wrote:
> On Thu, 2021-02-25 at 17:48 -0800, Dave Crocker wrote:
>> On 2/25/2021 1:05 PM, Kjetil Torgrim Homme wrote:
>> > Speaking of whitespace, the grammar uses LWSP = *(WSP /
>> > CRLF WSP) This is IMHO at odds with "The content of this
>> > part is restricted to single line of emoji." Why allow
>> > CRLF if only a single line is allowed? Why restrict to a
>> > single line?
>>
>> Good catch. It should be WSP and not LWSP.
>
> Thanks! This has the added benefit that emojis are required
> to be whitespace separated (whereas LWSP has an empty
> expansion). Well, I call it a benefit because I think it
> gives a better hint about each emoji being a separate
> reaction, for MUAs which want to do tallies and summaries.
I don't want to spend a lot of time on this unless we are at the
point that people have concluded it is time to call off the Last
Call and start reviewing the document (something I, for one,
have absolutely not concluded), but...
(1) If support for tallies and summaries, presumably including a
message being sent out in bulk and having recipients respond to
it in a way that is equivalent to voting, is part of the
intended use of this feature, why is that not called out in the
document? Unless I have missed something even after reading it
several times in the last few days, he document now seems to be
focused on interpersonal communications and not such tasks,
tasks that might call for a review of security considerations
and other text.
In addition and at least equally important, unless either the
sending and receiving systems are dedicated application rather
than conventional MUAs, or the vocabularies are extremely
limited, any notion of tabulating responses requires begin able
to determine whether two emoji or emoji sequences are "the
same". That need to do comparison, especially in combination
with there now being unitary (since code point) emoji whose most
likely symbolic representations can also be represented by
non-trivial sequences involves hard problems, with the absence
of the normalization techniques that help with text making them
harder.
For one of the most obvious examples (although probably not the
most difficult one) consider whether (all of the words below
representing emoji and skipping over combining and qualifier
issues) are the following three and other permutations
equivalent?
man-woman-child
man-child-woman
woman-child-man
are any or all of them equivalent to the "family" emoji
character?
So, if the authors (or, now, anyone else) contemplate using
reactions in a way that would lend itself to tabulation or other
operations involving comparisons, text and a health warning are
needed for that too.
(2) When I see multi-grapheme (as distinct from
multi-code-point) emoji in contexts that might be described as
responses, among the most common ones I see are sequences
intended to convey
hug-hug-hug
using the :TEDDY-BEAR: emoji. But
hug hug hug
which the above change appears to force, could easily have
rather different semantics especially if used in conjunction
with other emoji characters, e.g., of a teddy bear picnic. And,
coming back the first point, if the spaces are allowed and
comparison testing is a requirement, are the spaces significant?
Both of the above lead us away from the simplicity and principle
of starting out as simple as possible and extending only when
the need is clear that Dave eloquently argued for and which I
strongly support. Much as I hate the idea, they also seem to me
to be taking us close to the line beyond which a new version of
the spec would represent a great enough change to require an
additional Last Call.
best,
john
--
last-call mailing list
last-call@xxxxxxxx
https://www.ietf.org/mailman/listinfo/last-call
