Re: inputting characters by hexadigit

[Chong Yidong]

Juri Linkov <address@hidden> writes:

>>>> From my experience using C-q to input a number was a pain in the echo area.
>>> IMHO, using the minibuffer is more convenient that reading a string
>>> with a sequence of `read-event'.  That's why I have doubts about moving
>>
>> Then use C-x 8 RET.  No need for us to choose.
>
> I installed a patch that changes `ucs-insert' to read Unicode names.
> Anyone could improve it further with a discussed multi-level
> completion mechanism.
>
> A patch below binds `ucs-insert' to `C-x 8 RET'.  I tried to bind RET
> to iso-transl-ctl-x-8-map in iso-transl.el, but then `C-x 8 RET' calls
> `ucs-insert' as a non-interactive function (i.e. ignoring its `interactive'
> declaration).  However, `ucs-insert' bound directly to `ctl-x-map' as below
> works without problems.

Look OK to me.

I'm curious, though: by default, when you use C-q to input a number, the
radix is octal; by changing read-quoted-char-radix, you can change this
to decimal or hexadecimal.  In contrast, ucs-insert always reads the
code in hexadecimal.  I don't use these commands myself, so I'm
wondering which is the preferred behavior (based on the defaults in
other applications, and so forth)?