> Er, by "permission" I meant it doesn't make the _implementation_
> non-compliant.
No, it doesn-t make cthe compiler non-complaint, but it makes your
code non complaint.
> And implementation-defined is not the same as undefined.
I perfectly know it, but in this case it is not exactly said
if it is implementation-defined or undefined, so I think
the second applies.
>> - Each header declares and defines only those
>> identifiers listed in its associated section: If the header includes
>> another header then it will break this rule.
>
> I think this is meant as a statement that strictly conforming programs
> may not rely on them defining anything else. Most of these identifiers
The sentence from the standard is: "Each header declares and defines
only those identifiers listed in its associated section". I think is
really clear the meaning of the word _only_. If it declares more
identifiers then it can break conforming codes that use the same
identifier. This is the reason why all the new keywords of the
language are added with _ followed by upper case, because it is a
reserved namespace and it is sure that it will not break any code.
For example I have a program that is strictly conforming, but it
cannot compile in gcc because it uses the identifier typeof, that gcc
uses, even when it is in the program namespace. Gcc is not a
conforming implementation (of course it respects this point if you use
-ansi).
You didn't mention anything about the other reference I posted here,
the book of Plauger. I supossed ppl knew about this book and the
author, but maybe it is not true. This is the main staff of the ansi
committee:
Chair
Jim Brodie
Vice-Chair
Thomas Plum Plum Hall Secretary
P. J. Plauger Whitesmiths, Ltd.
International Representative
P. J. Plauger Whitesmiths, Ltd.
Steve Hersee Lattice, Inc.
Vocabulary Representative
Andrew Johnson Prime Computer
Environment Subcommittee Chairs
Ralph Ryan Microsoft
Ralph Phraner Phraner Associates
Language Subcommittee Chair
Lawrence Rosler AT&T
Library Subcommittee Chair
P. J. Plauger Whitesmiths, Ltd.
Draft Redactor
David F. Prosser AT&T
Lawrence Rosler AT&T
Rationale Redactor
Randy Hudson Intermetrics, Inc.
You can see that he was vice-chair of the committee and
chair of Library Subcommittee, and he wrote two books about
the standard, that basically were a translation for humans of it,
so if he wrote that standard headers cannot include another
standard headers I suppose he was rigth.
I think this conversation begins to be a bit OT, if you want we can
continue it in comp.lang.c.
Regards,
