This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: [PATCH] C undefined behavior fix
- From: dewar at gnat dot com
- To: dewar at gnat dot com, torvalds at transmeta dot com
- Cc: gcc at gcc dot gnu dot org, rth at redhat dot com, trini at kernel dot crashing dot org
- Date: Sat, 5 Jan 2002 07:21:55 -0500 (EST)
- Subject: Re: [PATCH] C undefined behavior fix
<<I do try to wear both hats, partly because I have noticed that if I do
_not_ wear the language lawyer hat I will invariably be answered by
language lawyer answers.
>>
Ah ha! Not in this particular case. The notion of implementation defined
tells the language lawyers to buzz off, not wanted here, this is to be
resolved on a pragmatic basis.
I would *NEVER* have brought the standard into this discussion if you had
not tried to argue (erroneously) that the standard supported your position.
Yes, implementation defined means that something has to be documented, but
there is no definition of what documented means. As I pointed out earlier,
we could perfectly well say that the code of gcc *is* the documentation.
That's absurd from a pragmatic point of view, but the language lawyers
have no choice but to agree, since they did not specify what documentation
meant in their contract, and they get to leave the room!