Re: sharp and ss in Ada keywords like ACCESS

["Randy Brukardt" <randy@rrsoftware.com>]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 2970 bytes --]

"Dmitry A. Kazakov" <mailbox@dmitry-kazakov.de> wrote in message 
news:1jwwodd91xfc8$.1xbuqt6j4xh5$.dlg@40tude.net...
> On Tue, 11 Oct 2011 15:32:35 -0500, Randy Brukardt wrote:
>
>> I believe (but haven't checked carefully), that Unicode case folding 
>> never
>> treats Cyrillic and Latin characters as the same (even when it could).
>
> That is for sure, they are different code points.

That has nothing to do with. I think you don't understand very well how 
identifier equivalence is defined by Unicode (which is borrowed by Ada). 
Unicode defines something called "full case folding" (as well as "simple 
case folding"), it defines which characters are considered equivalent by 
case. Different code points can of course be equivalent (think 'J' and 'j' 
in the Latin characters, and also the similar Cyrillic characters).

>> So this problem would not come up in Ada.
>
> Depends on what is considered problematic.
>
> � is not ss and for that matter, it is not ? (beta). Their capital 
> letters,
> if exist (� does not have capital case) are different.

Actually, "full case folding" says that "�" == "ss" == "SS" etc. (That's why 
we decided that we had to diverge from the Unicode recommendations, as the 
above would be incompatible with Ada 95 programs. Thus Ada 2012 uses "simple 
case folding", which does not map different length sequences as equivalent.)

> If Ada wished to introduce some rules of equivalence for Central European
> languages, like �=ss, �=ae, �=oe, and Eastern European languages, like 
> a=?,
> ?=e, and who knows what rules in other languages exist, that would be
> hopeless.

If Unicode wanted to include such equivalences, Ada would have them too. 
It's not at all hopeless - we're simply going to do what ever Unicode 
recommends. (In this specific case, I believe that Unicode recommends not 
have equivalence.)

> But without them, the programmer could have *all* Ada keywords as
> identifiers, by replacing appropriate Latin letters with Cyrillic ones.

Yup. And so what? It's the price you pay for allowing Cyrillic characters in 
identifiers. And that is a choice completely out of our hands -- it is 
mandated by JTC 1 and SC 22. We couldn't have an Ada Standard at all without 
allowing this.

> Furthermore, identifiers looking perfectly same will be different and 
> there
> is a huge number of homonyms of practically *each* reasonable identifier. 
> I
> see it in breach with basic Ada design.

As mentioned, it is required as a part of all programming language 
standards. Is this a bad idea? Sure, I agree with you on that, but we have 
no choice.

J-P's tool or even a compiler-switch can prevent problematic identifiers. 
We're not allowed to if we want to be an International Standard (emphasis on 
"International"). [Or just use Janus/Ada; it's unlikely that I'll mess with 
the pain of these identifiers unless some customer demands it with $$$.]

                                       Randy.