Re: Seeking for papers about tagged types vs access to subprograms

["Dmitry A. Kazakov" <mailbox@dmitry-kazakov.de>]

On Wed, 15 May 2013 06:24:49 +0200, Yannick Duchêne (Hibou57) wrote:

> Le Tue, 14 May 2013 21:56:14 +0200, Robert A Duff  
> <bobduff@shell01.theworld.com> a écrit:
>> Furthermore, lots of programs do NOT want that two-level structure.
>> They just want a sequence of characters, with one (not two!)
>> particular character that means "line break".  They want to have
>> code like:
>>
>>     ... loop
>>         case Current_Char is
>>             ...
> 
> That's a corner case, and why Ada should do pre‑tokenization for this  
> single case?

It is not a tokenization. You turned abstraction upside down. The text file
is a sequence of lines. A line is a sequence of characters. Period.

If you have a broken OS, like Linux/Windows is, you cannot have this
abstraction. Instead you emulate it as a lower-level stream of octets with
escape sequences used to simulate lines. What is wrong with that? It
conflates representation with the interface.

> If end‑of‑line is made of two characters in some  
> environments, then the end of line token is not a character, or else it's  
> a character in a kind of internal character set (what most people surely  
> do, indeed, and that's out of the standard).

It is not how it was under VMS. It did not have such stuff in record
oriented files. Such files had each line kept with the character count. It
was not a stream of characters. Though under VMS you could read it as a
stream of octets if you wanted.

-- 
Regards,
Dmitry A. Kazakov
http://www.dmitry-kazakov.de