On Sun, 28 Jul 2002, Tal Dayan wrote:

>> Of course some will also point out that
>> it's not true, pure, 100% ANSI C, and they're quite right.
>
>The main point is not it is not a pure ANSI C but that it lakes
>elementary features that are available in most commercial C compilers
>written in the last 10 years or more.
>
>For example, generating an error when a function that is declared as 'int
>foo()' does not return any value. Or, that it does not complain when you
>call a function with a pointer instead of a structure.

imho the compiler's job is to compile whatever the programmer throws at
it. Syntax checking is another tool's job, for example (c)lint and many
others. I have not so fond memories of jumping through hoops on fire to
make a Microsoft compiler and assembler do what I wanted (non-standard
boot code for embedded). It insisted on 'strong typing' in assembler and
used segment registers that did not exist in my environment. Yeah, right.
The Borland toolchains dealt with it directly without trying hard. I don't
want the compiler to check what I pass it as a structure ...

Peter

--
http://www.piclist.com hint: The PICList is archived three different
ways.  See http://www.piclist.com/#archives for details.