Re: Array new followed by non-array delete - request for experience
<jkherciueh@gmx.net> schrieb im Newsbeitrag news:fe6b6m$mn0$1@aioe.org...
Matthias Hofmann wrote:
Dereferencing a null pointer is undefined behavior and so
is relying on std::string being contiguous. While I will avoid the first
at
all costs, I take the liberty of doing the later in my code (in fact,
sometimes I go as far as assuming that it is safe to write \0 behind the
string, which I think will remain undefined even with the next revision of
the standard).
What do you mean by relying on std::string being contigous? 21.3/2 says
that
"the iterators supported by basic_string are random access iterators", and
as std::string is nothing but a typedef to std::basic_string<char>, the
same
must be true for the former. And as far as I understand things, random
access interators imply that the elements of the sequence are contigous in
memory.
And what's the deal with writing '\0' behind the string? Where is "behind
the string", do you mean the element whose index is basic_string::size()?
According to 21.3.6/1, there already is "a null character specified by
charT()" at that position. And the type of the value returned from
basic_char::c_str() is const charT*, so as a matter of course, you cannot
use it to modify any of the elements of the string.
Whether rational people get scared by undefined behavior depends on the
consequences of which they are aware.
So for rational people the term "undefined" is too... well, "undefined" to
be scary? They need a more "defined" concept of "undefined"?
If you show them that a certain piece
of code not only _can_ do something unexpected but is likely to _do_
something unexpected, they will be more likely to change the code.
The mere thought of having to explain to someone what may happen if
anything
can happen makes me a little upset! Isn't it enough to say the the program
may crash, with all the consequences a crash may have?
If the question is only about calling scalar delete on a pointer obtained
from array new, then write an email to your compiler vendor and ask him
what
the behaviour will be. If he says it will be fine, then you can consider it
to be a language extension. Note, for example, that polymorphic array
deletion is also undefined behaviour, but it does seem to work on
Microsoft's compilers:
"Although, strictly speaking, polymorphic array delete is undefined
behavior, we had several customer requests to implement it anyway.
Therefore, in MSC++, this is implemented by yet another synthesized virtual
destructor helper function, the so-called "vector delete destructor," which
(since it is customized for a particular class, such as WW) has no
difficulty iterating through the array elements (in reverse order), calling
the appropriate destructor for each."
(From the MSDN Library article "C++: Under the Hood" by Jan Gray from March
1994, which can be found here:
http://msdn.microsoft.com/archive/default.asp?url=/archive/en-us/dnarvc/html/jangrayhood.asp
)
--
Matthias Hofmann
Anvil-Soft, CEO
http://www.anvil-soft.com - The Creators of Toilet Tycoon
http://www.anvil-soft.de - Die Macher des Klomanagers
[ See http://www.gotw.ca/resources/clcm.htm for info about ]
[ comp.lang.c++.moderated. First time posters: Do this! ]