Removing set-unibyte-charset

Removing set-unibyte-charset

[Stefan Monnier]

I'd like to remove set-unibyte-charset.
Any objection?


        Stefan

Re: Removing set-unibyte-charset

[Kenichi Handa]

In article <jwvablgu6pw.fsf-monnier+emacs@gnu.org>, Stefan Monnier <monnier@iro.umontreal.ca> writes:

> I'd like to remove set-unibyte-charset.
> Any objection?

I don't object.  But, why now?  Are you going to change the
behaviour of unibyte<->multibyte conversion?

---
Kenichi Handa
handa@ni.aist.go.jp

Re: Removing set-unibyte-charset

[Stefan Monnier]

>> I'd like to remove set-unibyte-charset.
>> Any objection?

> I don't object.  But, why now?

The switch to 23.1 seems like a good time for it.

> Are you going to change the
> behaviour of unibyte<->multibyte conversion?

Clearly that will change it, indeed.
E.g. string-make-multibyte == string-to-multibyte after that removal,
for example.


        Stefan

Re: Removing set-unibyte-charset

[Kenichi Handa]

In article <jwvablfn8yb.fsf-monnier+emacs@gnu.org>, Stefan Monnier <monnier@iro.umontreal.ca> writes:

> > Are you going to change the
> > behaviour of unibyte<->multibyte conversion?

> Clearly that will change it, indeed.
> E.g. string-make-multibyte == string-to-multibyte after that removal,
> for example.

Ok, I see.  It's a big overhaul.  :-)

---
Kenichi Handa
handa@ni.aist.go.jp

Re: Removing set-unibyte-charset

[Stefan Monnier]

>> > Are you going to change the
>> > behaviour of unibyte<->multibyte conversion?

>> Clearly that will change it, indeed.
>> E.g. string-make-multibyte == string-to-multibyte after that removal,
>> for example.

> Ok, I see.  It's a big overhaul.  :-)

What do you mean by a big overhaul?  The first step is nothing more than
removing Fset_unibyte_charset (and removing the few calls to it).

After that, many more changes can be applied, but these are nothing more
than optimizations (e.g. remove unibyte-charset since it's constant,
...).


        Stefan

Re: Removing set-unibyte-charset

[Kenichi Handa]

In article <jwv4pbmbgc2.fsf-monnier+emacs@gnu.org>, Stefan Monnier <monnier@iro.umontreal.ca> writes:

>>> > Are you going to change the
>>> > behaviour of unibyte<->multibyte conversion?

>>> Clearly that will change it, indeed.
>>> E.g. string-make-multibyte == string-to-multibyte after that removal,
>>> for example.

> > Ok, I see.  It's a big overhaul.  :-)

> What do you mean by a big overhaul?  The first step is nothing more than
> removing Fset_unibyte_charset (and removing the few calls to it).

Yes, but...

> After that, many more changes can be applied, but these are nothing more
> than optimizations (e.g. remove unibyte-charset since it's constant,
> ...).

I've thought that you are going to change all operations
that convert unibyte char/string/buffer-contents to
multibyte (insert, concat, search, re-search, etc).
Currently they do "make"-conversion; i.e. convert each
unibyte character to multibyte by using unibyte-charset.
Changing all of them to "to"-conversion is a big (but not
that difficult) work, I think.

---
Kenichi Handa
handa@ni.aist.go.jp

Re: Removing set-unibyte-charset

[Stefan Monnier]

>> After that, many more changes can be applied, but these are nothing more
>> than optimizations (e.g. remove unibyte-charset since it's constant,
>> ...).

> I've thought that you are going to change all operations
> that convert unibyte char/string/buffer-contents to
> multibyte (insert, concat, search, re-search, etc).
> Currently they do "make"-conversion; i.e. convert each
> unibyte character to multibyte by using unibyte-charset.
> Changing all of them to "to"-conversion is a big (but not
> that difficult) work, I think.

It's a tiny amount of work: just removing all calls to
set-unibyte-charset does the trick.  Admittedly, the subsequent
optimizations that become possible represent a fairly large amount of
changes, but as you say it's not difficult, and it can be done at
a leisurely pace.


        Stefan