Re: gcc/gcc/cp ChangeLog dump.c

[Nathanael Nerode <neroden at twcny dot rr dot com>]

Sorry, just realized, reading a different mail of yours that you made a 
major, and very confusing typo.

Gaby:
>But the assertion that the C version ought to return a bool is not
>true.

This is the exact opposite of what you mean, isn't it?  You mean:

>But the assertion that the C version ought not to return a bool is not
                                            ^^^
>true.

In the other message:

Gaby:
>That is not true.  The C version *can* safely return bool.
Buuuut... the C version, and the generic langhooks table, have to be K&R 
compatible because they're built during stage 1 using the bootstrap 
compiler.  Doesn't this indicate that they can't return bool (a.k.a. 
char)?  Or does (our version of) bool work as a return type in K&R 
compilers, in which case the boolification change has *nothing* to do 
with ANSIfication?  I'm confused by what you're saying.

>Me:
>| Boolification is conceptually separate from declaration
>| and definition style -- and a *lot* more bug-prone.
>
>No.  Again that is untrue.

Whether it's more bug-prone is a matter of opinion; I think it is; I 
assume you think it isn't.  That it is conceptually separate is *not* a 
matter of opinion.  It *is* conceptually separate.