From: "Rafał Maj" <r_maj AT poczta DOT onet DOT pl>
Newsgroups: comp.os.msdos.djgpp
Subject: Odp: FINALY: delete/delete[] with build-in-types/user-classes
Date: Sun, 17 Sep 2000 14:48:24 +0200
Organization: Academic Computer Center CYFRONET AGH
Lines: 85
Message-ID: <8q2egd$o5j$1@info.cyf-kr.edu.pl>
References: <8q1tn2$iv6$1 AT info DOT cyf-kr DOT edu DOT pl> <i3zEOX02JyFsEFt0MbK72k1ScoTo AT 4ax DOT com>
NNTP-Posting-Host: d-94-53-17.cyfronet.krakow.pl
X-Trace: info.cyf-kr.edu.pl 969194829 24755 149.156.1.177 (17 Sep 2000 12:47:09 GMT)
X-Complaints-To: news AT cyf-kr DOT edu DOT pl
NNTP-Posting-Date: 17 Sep 2000 12:47:09 GMT
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 5.00.2615.200
X-MimeOLE: Produced By Microsoft MimeOLE V5.00.2615.200
To: djgpp AT delorie DOT com
DJ-Gateway: from newsgroup comp.os.msdos.djgpp
Reply-To: djgpp AT delorie DOT com

Thank You very much :-)))

Użytkownik Jack Klein <jackklein AT spamcop DOT net> w wiadomooci do grup
dyskusyjnych napisał:i3zEOX02JyFsEFt0MbK72k1ScoTo AT 4ax DOT com...
> On Sun, 17 Sep 2000 10:00:28 +0200, "Rafał Maj" <r_maj AT poczta DOT onet DOT pl>
> wrote in comp.os.msdos.djgpp:
>
> > Hi,
> > because my last question about delete / delete[] caused big discusion
with
> > meany diffrent opinions, I wan't to check, if finaly I understand it...
> > (probably not ;)
> > 1) "s" is a pointer to array, maked like:  myclass* s = new
myclass[100];
> >    a)  delete []s    will not only free memory, but alsow call
destructor
> > for each of 100 objects "myclass"
>
> Yes, that is correct.
>
> >    b) using delete s;  is probably wrong, because none destructor will
be
> > called. But all memory 100*sizeof(myclass) will be freed ?
>
> Maybe, maybe not.  The C++ language standard states that items created
> with new[] be destroyed by delete[], and those with new by delete.
> Combining new[] with delete or new with delete[] is undefined
> behavior, which means anything can happen.  Even if this happens to
> work with one particular, it might not with the next one.
>
> > 2) with   char *s = new char[100]  I can use :
> >   a) delete []s;  but it isn't necessarly, when char doesn't have any
> > destructor
> >   b) delete s;  is good
>
> No, delete after new[] is not good.  It is illegal code, it is a bug,
> it is not valid C++.  Even if it seems to work.
>
> > 3) and what with   char* s = malloc(sizeof(char)*100); ?
>
> Actually you can't even compile this as C++, although you can as C.
> In C++ you can't assign a void * to another type of pointer without a
> cast.  And you never, never need to multiply anything by sizeof(char),
> since sizeof(char) is 1 by definition, and always will be.  The safest
> way to use malloc() is:
>
> char *s = (char *)malloc(100 * sizeof *s);
>
> In C, don't use the (char *) cast.
>
> >    a) free - typical
>
> Yes, or realloc() with a size of 0.
>
> >    b) can I use delete ?
>
> No.
>
> >    c) can I use delete [] ?
>
> No.
>
> > 4) because strdup() uses malloc,   char *s=strdup(S); should be
"deleted"
> > same way as in question 3
> >
> > Many Thanks,
> > Rafal
>
> There are three pairs of memory management functions, and you should
> never, NEVER, mix them:
>
> new & delete
> new[] & delete[]
> malloc() or calloc() & free()
>
> You can use all of these different methods in the same program for
> different allocations, but for any given memory block you must free it
> with the corresponding type you used to allocate it.
>
> Jack Klein
> --
> Home: http://jackklein.home.att.net