Re: a great C++ book?

[kaih@khms.westfalen.de (Kai Henningsen)]

jamagallon@able.es (J.A. Magallon)  wrote on 02.01.02 in <20020102013411.A5968@werewolf.able.es>:

> On 20020101 Larry McVoy wrote:
> >
> >Makes you wonder what would happen if someone tried to design a
> >minimalistic C++, call it the "M programming language", have be close
> >to C with the minimal useful parts of C++ included.
> >
>
> There are specs for something called 'Embedded C++'. You can run it on
> a cell phone, so it looks like little bloated...

Wrong metric, unless you mean you can run the *compiler* on the cell  
phone.

c99.pdf:   1412026 bytes
c++98.pdf: 2860601 bytes

And remember that interactions between features go up exponentially.

I think one of the worst design decisions for C++ (hindsight, of course)  
was to keep compatibility with C. The other bad one was to accept too many  
new features.

The Modula-3 standard (that's an OO variant of Modula-2) had as a language  
design goal that the language description should not take more than 50  
pages. They overshot that, but it still was much less than 100 (somewhat  
less precise than C/C++, admittedly). As a result, that language (and  
consequently, programs in that language) is *much* easier to understand  
than C++.

That doesn't mean I'm in love with Modula-3. Actually, I don't even use  
it, and I do think (hindsight again) some of the design decisions were  
unfortunate. But I do think keeping an eye on the sheer mass of the spec  
is a good idea.

MfG Kai