Re: tree_node using std::vector

From:

=?ISO-8859-1?Q?Daniel_Kr=FCgler?= <daniel.kruegler@googlemail.com>

Newsgroups:

comp.lang.c++.moderated

Date:

Mon, 19 May 2008 06:27:31 CST

Message-ID:

<b58bf377-6099-4d8d-97d0-ca7dda26e51c@p25g2000hsf.googlegroups.com>

On 19 Mai, 00:48, Maik Beckmann <maikbeckm...@gmx.de> wrote:

Daniel Kr?gler wrote:

On 18 Mai, 07:28, Maik Beckmann <maikbeckm...@gmx.de> wrote:

Ok, its undefined. This means if one wants to use this "undefined" code
it should to be checked via the build system.

Mind to explain in more detail? What has a build-system
to do with undefined behavior? Build-systems are similarly
not specified by the standard and I do not see a relation
between both. Your observation that gcc-under-given-configuration
produces a diagnostic is not required, but one possible form
of outcome of undefined behavior. Specifically, any lack of
diagnostics does not nominate this as "defined" code.

I think this
struct Foo { vector<Foo> m_Foos; };
fails to compile or works proper. Before I can use a compiler I have to
check if this code is compiled which can be scripted by a build system.

I don't think that your assumption is correct, please see my
answer to Greg Herlihy, where I give two exemplary counter-
examples, which are not required to be diagnosed and which
are clearly specified to have undefined behavior.

More experienced people than me will probably tell you
that some earlier versions of gcc started a simple computer
game, when the compiler (or the executable - I don't know in
detail) stumbled across some form of undefined behavior in
the code.

Was it about this specific case?

I just wanted to give a more humorous example of
undefined behavior. Typical answers of others are
that your program might send an e-mail to your boss.

Greetings from Bremen,

Daniel Kr?gler

--
[ See http://www.gotw.ca/resources/clcm.htm for info about ]
[ comp.lang.c++.moderated. First time posters: Do this! ]