etc/nxml/*.el duplicate lisp/international/uni-name.el?

etc/nxml/*.el duplicate lisp/international/uni-name.el?

[Kan-Ru Chen]

It seems etc/nxml/*.el, some unicode name to code point tables,
are only used by `nxml-insert-named-char'. It looks like a variant of
`ucs-insert' while outputs NCR instead of real character.

Should nxml use `read-char-by-name' so we could remove a bunch of
duplicated data?

Re: etc/nxml/*.el duplicate lisp/international/uni-name.el?

[Chong Yidong]

Kan-Ru Chen <kanru@kanru.info> writes:

> It seems etc/nxml/*.el, some unicode name to code point tables,
> are only used by `nxml-insert-named-char'. It looks like a variant of
> `ucs-insert' while outputs NCR instead of real character.
>
> Should nxml use `read-char-by-name' so we could remove a bunch of
> duplicated data?

The tables are also required by nxml-get-char-name, so we would also
need a generic replacement for that.  Help welcome.

Re: etc/nxml/*.el duplicate lisp/international/uni-name.el?

[Kan-Ru Chen]

Chong Yidong <cyd@stupidchicken.com> writes:

> Kan-Ru Chen <kanru@kanru.info> writes:
>
>> It seems etc/nxml/*.el, some unicode name to code point tables,
>> are only used by `nxml-insert-named-char'. It looks like a variant of
>> `ucs-insert' while outputs NCR instead of real character.
>>
>> Should nxml use `read-char-by-name' so we could remove a bunch of
>> duplicated data?
>
> The tables are also required by nxml-get-char-name, so we would also
> need a generic replacement for that.  Help welcome.

I'd like to help!

But I am wondering what function/variable should be considered public
interface to be kept compatible. For example, currently nxml allow
customize nxml-enabled-unicode-blocks, so I have to keep it, but what
about other defuns/defvars?

- Newbie to Emacs internals

Re: etc/nxml/*.el duplicate lisp/international/uni-name.el?

[Stefan Monnier]

>>> It seems etc/nxml/*.el, some unicode name to code point tables,
>>> are only used by `nxml-insert-named-char'. It looks like a variant of
>>> `ucs-insert' while outputs NCR instead of real character.
>>> Should nxml use `read-char-by-name' so we could remove a bunch of
>>> duplicated data?
>> The tables are also required by nxml-get-char-name, so we would also
>> need a generic replacement for that.  Help welcome.

> But I am wondering what function/variable should be considered public
> interface to be kept compatible.  For example, currently nxml allow
> customize nxml-enabled-unicode-blocks, so I have to keep it, but what
> about other defuns/defvars?

We generally don't know.  So the best you can do is to make a wild guess
(tho sometimes an internet search can help figure out if something is
used by external packages), then post it here.  If noone objects here,
then it may/will get installed and maybe later we'll get complaints
about this change, at which point we may try and adjust the
implementation or accept the problem as an unfortunate consequence.

In the case of nxml-enabled-unicode-blocks, it may not be necessary to
preserve this customization, depending on what impact it may have.


        Stefan

Re: etc/nxml/*.el duplicate lisp/international/uni-name.el?

[Kan-Ru Chen]

Stefan Monnier <monnier@IRO.UMontreal.CA> writes:

> We generally don't know.  So the best you can do is to make a wild guess
> (tho sometimes an internet search can help figure out if something is
> used by external packages), then post it here.  If noone objects here,
> then it may/will get installed and maybe later we'll get complaints
> about this change, at which point we may try and adjust the
> implementation or accept the problem as an unfortunate consequence.
>
> In the case of nxml-enabled-unicode-blocks, it may not be necessary to
> preserve this customization, depending on what impact it may have.

Thank you. Now I can go ahead witch confidence.