I should mention, here, that I am ignoring the "fread/fwrite" part of
your proposal, for now, because your motivating examples did not
involve the use of fread nor fwrite.

-- 
Raul

On Sun, Mar 20, 2022 at 11:36 AM Raul Miller <[email protected]> wrote:
>
> On Sun, Mar 20, 2022 at 2:38 AM Elijah Stone <[email protected]> wrote:
> > > Deprecation based on something which has not been implemented is bad
> > > news.
> >
> > What are you getting at?  That I did not include a deprecation plan in my
> > proposal?  I would like to first establish what the ideal semantics should
> > _be_.  Getting there is secondary.
>
> Getting there, from my perspective, should be thought of as critical
> for any valid proposal.
>
> > > I think a point has been lost here on why getting rid of u: would not
> > > change anything about the initial example in this thread:
> >
> > I think I addressed this; maybe I was not clear enough.  Removing u: alone
> > is not sufficient to solve the problem; but it is necessary, and my
> > proposal in its entirety remedies the problem.
>
> From my perspective, the problem is inherent in the *unicode*
> specification. It's not a problem in the J *language*. (J is based on
> ascii, not unicode. J has added some minimal support for unicode,
> which is also based on ascii.)
>
> So, from a language perspective, we could obscure the issues. But the
> underlying "discontinuities" would remain.
>
> Now... your proposal -- which would have the unicode latin-1 code
> plane be the default interpretation for J's literal datatype when (for
> example) promoting from literal to unicode4 -- does sound attractive.
> And, that is similar to historical J (where we also treated those
> characters as representing the latin-1 code plane, instead of utf-8),
> and we moved away from that because of the ubiquity of utf-8 in OS
> interfaces.
>
> (Making J be a unicode language instead of an ascii language is
> something I remember proposing, years ago. But the work involved is
> overwhelmingly open-ended. (If you want a unicode language, I think
> raku is a great start. And, since the raku design proposals explicitly
> include the capabilities needed to have J as a subset of the language,
> I imagine that eventually you could load J packages directly into
> raku, perhaps even with version support so that deprecated J graphics
> code could function there. Eventually. (Probably with some performance
> penalties or possibly even performance improvements, though that issue
> is less clear.)))
>
> Anyways, ...
>
> Thinking in terms of implementation, a key concept embedded in your
> proposal is that we remove the ability to talk directly with OS
> interfaces which expect utf-8. But that seems to conflict, directly,
> with the principles you proposed as the motivation for your proposal.
>
> Why would you do that?
>
> Thanks,
>
> --
> Raul
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm

Reply via email to