[Jprogramming] RFC: unicode

[Elijah Stone]

On Sat, 19 Mar 2022, Raul Miller wrote:

>> I do not need to carefully track which words are integers and which are 
>> floats.
>
> This is sometimes necessary. I've worked with examples where 
> distinguishing between numeric types has been a critical issue. I have 
> even encountered examples where the distinction between integer and 
> floating point was key.

Good point!  I bumped into one instance of this recently: x^y for 
nonnegative integers x and y is a float.

I think the broader point still applies, though.  Take booleans and 
integers, say, rather than floats and integers.


> https://code.jsoftware.com/wiki/Essays/Deprecation

I think a deprecation period would probably be a good idea.


> ": does nothing to content which is already textual

Per the dictionary:

> ": converts literal2 and literal4 to U8 encoded 1-byte char

And experimentally:

    ]x=. 10 u: 243 244
óô
    datatype x
unicode4
    #x
2
    ":x
óô
    datatype ":x
literal
    #":x
4

Not specified is whether literal2 is interpreted as ucs-2 or utf-16. 
Experimentally, it is utf-16.


> The fill mechanism only presumes that the data is in a form which can be 
> filled with the default fill for that type. If that's not what you want, 
> you should avoid operations which add fill. For example, you may want to 
> use an expression like:
>
>   ; verb each sequence

I don't understand the significance of this.


> Generally speaking, if you want an unambiguous representation of your 
> data, you should use something like {{ 5!:5<'y' }} rather than ":

I don't need unambiguous.  I'll take non-obfuscatory.  And, as mentioned, 
the behaviour of ": here is inconsistent with other primitives.


>> z is a perfectly good, valid utf-32 sequence.
>
> Hmm... I suppose you are correct. Which means the problem here is that 
> it is not being displayed correctly.

The display seems correct to me.