Emacs release schedule and Akihito's abdication

Emacs release schedule and Akihito's abdication

[Paul Eggert]

Should we plan for a new release around March 2019, if only a point release? 
Otherwise, a lot of Japanese text will stop working, due to the expected 
abdication of Emperor Akihito in April 2019, and the use of a new character in 
Japanese-format dates. This differs from ordinary Unicode updates, since the new 
character is likely to be quite common pretty quickly, at least in Japanese 
text. The new character will not be announced until late February 2019, 
unfortunately. See:

Whistler K. Unicode 12.1 Planning Considerations. 2018-07-16. 
https://www.unicode.org/L2/L2018/18220-u121planning.txt

Hern A. Big tech warns of 'Japan's millennium bug' ahead of Akihito's 
abdication. The Guardian. 2018-07-25. 
https://www.theguardian.com/technology/2018/jul/25/big-tech-warns-japan-millennium-bug-y2k-emperor-akihito-abdication

Re: Emacs release schedule and Akihito's abdication

[John Wiegley]

>>>>> "PE" == Paul Eggert <eggert@cs.ucla.edu> writes:

PE> Should we plan for a new release around March 2019, if only a point
PE> release? Otherwise, a lot of Japanese text will stop working, due to the
PE> expected abdication of Emperor Akihito in April 2019, and the use of a new
PE> character in Japanese-format dates. This differs from ordinary Unicode
PE> updates, since the new character is likely to be quite common pretty
PE> quickly, at least in Japanese text. The new character will not be
PE> announced until late February 2019, unfortunately.

I think a patch to the release branch would be reasonable for that purpose; I
wonder how long it will take for that release to make it into all channels.

-- 
John Wiegley                  GPG fingerprint = 4710 CF98 AF9B 327B B80F
http://newartisans.com                          60E1 46C4 BD1A 7AC1 4BA2

Re: Emacs release schedule and Akihito's abdication

[Eli Zaretskii]

> From: Paul Eggert <eggert@cs.ucla.edu>
> Date: Fri, 27 Jul 2018 13:20:38 -0700
> 
> Should we plan for a new release around March 2019, if only a point release? 

Are you asking about Emacs 27.1 or about Emacs 26.2?  I hope the
latter will be released sooner than March 2019.

Incorporating a new version of Unicode is not a large job, especially
if the only change is addition of a single character.

Re: Emacs release schedule and Akihito's abdication

[Paul Eggert]

Eli Zaretskii wrote:
> Are you asking about Emacs 27.1 or about Emacs 26.2?

Whatever the current release is at the time, I expect. The question is more 
"will we be prepared to cut a release quickly" than predicting an exact release 
number now. I assume the answer is "yes", and am mostly bringing up the issue 
now as a heads-up.

Re: Emacs release schedule and Akihito's abdication

[Clément Pit-Claudel]

On 2018-07-27 16:20, Paul Eggert wrote:
> Should we plan for a new release around March 2019, if only a point release? Otherwise, a lot of Japanese text will stop working, due to the expected abdication of Emperor Akihito in April 2019, and the use of a new character in Japanese-format dates.

Thanks for highlighting the issue, Paul.

I hope you can indulge a naive question :)  Can you clarify how Japanese text would break?  My quite-possibly-incorrect mental model is that Emacs finds appropriate characters in fonts by codepoint, so assuming installed fonts support it, shouldn't Emacs should be able to display the new character just fine?

Of course, the output of C-u x = would likely be broken or partial, but would text display be broken too?

Thanks!
Clément.

Re: Emacs release schedule and Akihito's abdication

[Paul Eggert]

Clément Pit-Claudel wrote:
> Can you clarify how Japanese text would break?

Sorry, I don't know exactly. One thing is that the regular expression 
[[:alpha:]] wouldn't match the new character even though it'll be alphabetic. 
Although I expect more issues will arise, not being a heavy-duty Japanese user I 
don't know what they'll be.

 From the Unicode point of view, the issue is not merely assigning a code point 
for the new character (that's already done: it'll be U+32FF), it's assigning 
compatibility decompositions, something that won't be known until the new 
emperor chooses his name. For example, the current emperor's name ㍻ (U+337B) is 
a composition of 平成 (U+5E73, U+6210). Because decompositions must be cast in 
stone before the code point is officially issued, Unicode won't update its 
tables until it knows the actual name and how it's composed. And because it 
won't update its tables, I expect that Emacs (and other software apps) will wait 
until it does, which means that older programs may well mishandle the new character.

The new emperor could help out by choosing a name now, but Japanese imperial 
protocol long predates Unicode and I expect they're not gonna change how they do 
things just because some young whippersnapper computer dudes ask them to.

Also, suppose the new emperor decides on a single-character name that's already 
in Unicode; presumably U+32FF will not be assigned to the name after all. I 
think this unlikely, though, as getting a character of your own is one of the 
perks of being emperor.

There are other possibilities, including the unexpected death of an emperor 
(things get pretty complicated for dates, but I expect for text it's merely just 
keeping a catalog of all the names).

Re: Emacs release schedule and Akihito's abdication

[Clément Pit-Claudel]

On 2018-07-27 19:41, Paul Eggert wrote:
> Clément Pit-Claudel wrote:
>> Can you clarify how Japanese text would break?
> 
> Sorry, I don't know exactly. One thing is that the regular expression [[:alpha:]] wouldn't match the new character even though it'll be alphabetic. Although I expect more issues will arise, not being a heavy-duty Japanese user I don't know what they'll be.

Thanks.  Maybe someone else on the list will have further insight :)

> From the Unicode point of view, the issue is not merely assigning a code point for the new character (that's already done: it'll be U+32FF)

Indeed (I read your two links with great interest, and these issues were what I alluded to in mentioning C-u x =).  The point about [[:alpha:]] is a good one.  If it's the only issue, maybe we could save an extra release by ensuring that [:alpha:] includes U+32FF (we don't need to know what U+32FF will look like to add it to the appropriate regex class).

As I mentioned earlier, this is mostly curiosity :)
Clément.

Re: Emacs release schedule and Akihito's abdication

[Eli Zaretskii]

> From: Clément Pit-Claudel <cpitclaudel@gmail.com>
> Date: Sat, 28 Jul 2018 01:00:45 -0400
> 
> On 2018-07-27 19:41, Paul Eggert wrote:
> > Clément Pit-Claudel wrote:
> >> Can you clarify how Japanese text would break?
> > 
> > Sorry, I don't know exactly. One thing is that the regular expression [[:alpha:]] wouldn't match the new character even though it'll be alphabetic. Although I expect more issues will arise, not being a heavy-duty Japanese user I don't know what they'll be.
> 
> Thanks.  Maybe someone else on the list will have further insight :)

I'm not sure it is worth the hassle, but one might go through
admin/notes/Unicode, and consider whether any of the stuff we do there
when importing a new Unicode version may or may not be affected by
this issue.

Regardless, we always try to comply to the latest version of the
Unicode standard when we release Emacs, so if a draft of Unicode will
have existed with the new character by the time we make the release,
we could use that.

Re: Emacs release schedule and Akihito's abdication

[Van L]

> I hope you can indulge a naive question :)

Are there any memories from 1989 
at the last era change in Japan?

Re: Emacs release schedule and Akihito's abdication

[Paul Eggert]

Van L wrote:
> Are there any memories from 1989 at the last era change in Japan?

That predates Unicode. Although I don't have personal memories of the 
transition, I suspect that back then instead of ㍻ (U+337B SQUARE ERA NAME 
HEISEI) people just wrote the two characters 平成 (U+5E73, U+6210) as both of the 
latter are in JIS X 0208-1983. Most likely a similar substitute will be 
available next year, though we won't know for sure until late February at the 
earliest.

In my limited experience the two-character spelling is ubiquitous even today, 
and the special single character is used mostly in official documents.