Another solution is to transfer ownership of the handle to the new object.
E.g.: // Note that this requires the m_sqlite variable to be mutuable (for
modification of const-object).DatabaseRecord::DatabaseRecord(const
DatabaseRecord &rhs){ m_sqlite = rhs.m_sqlite; ...
rhs.m_sqlite = 0;}
> Date: Fri, 5 Aug 2011 20:08:16 +0800
> From: [email protected]
> To: [email protected]
> Subject: Re: [sqlite] sqlite-users Digest, Vol 44, Issue 4
>
> >
> > ------------------------------
> >
> > Message: 4
> > Date: Wed, 3 Aug 2011 16:25:13 +0200
> > From: Stephan Beal <[email protected]>
> > Subject: Re: [sqlite] Deep copy of 2 sqlite3*?
> > To: General Discussion of SQLite Database <[email protected]>
> > Message-ID:
> > <CAKd4nAj==+ga_dhfok6b8rpygfzqubkxqpsy3q37gtaz_7l...@mail.gmail.com
> > >
> > Content-Type: text/plain; charset=ISO-8859-1
> >
> > On Wed, Aug 3, 2011 at 12:56 PM, Teg <[email protected]> wrote:
> >
> > > I'd suggest either not doing it (prevent the copy constructor from
> > > compiling by declaring it private) or doing it by passing ownership of
> > > the connection around as if it was a socket or file handle.
> > >
> >
> > +1 to that.
> >
> > It is "almost never" a good idea to implement copy ctors for classes which
> > refer to stream handles and such.
> >
> > The absolute simplest/sanest/safest way to implement this is to declare the
> > copy ctor and copy assignment operator as private and NOT implement them.
> >
> > --
> > ----- stephan beal
> > http://wanderinghorse.net/home/stephan/
> >
> >
> > ------------------------------
> >
> >
> > ------------------------------
> >
> > Message: 6
> > Date: Wed, 3 Aug 2011 17:17:24 +0200
> > From: Stephan Beal <[email protected]>
> > Subject: Re: [sqlite] Deep copy of 2 sqlite3*?
> > To: General Discussion of SQLite Database <[email protected]>
> > Message-ID:
> > <CAKd4nAifdjN_S651O4RkzNJXzyKr=sPvD9KiBK0k=dFZED=o...@mail.gmail.com
> > >
> > Content-Type: text/plain; charset=ISO-8859-1
> >
> > On Wed, Aug 3, 2011 at 4:20 AM, Nikki Tan <[email protected]> wrote:
> >
> > > Hi sqlite-users!
> > > It's just that I'm writing a copy constructor of my own mini sqlite
> > > wrapper, and I ended up having to do a deep copy of a sqlite3* points to
> > a
> > > SQLite connection. And I'm just wondering is it okay that I just do it
> > with
> > > memcpy(), by digging into the code for the definition of struct sqlite3
> > > and
> > > count the total bytes of this struct?
> > >
> >
> >
> > Another point: memcpy() will NOT do what you want here because the sqlite3
> > structure dynamically allocates OTHER structures and points to them. A
> > memcpy() will copy the pointers but not their contents. i.e. you'll have 2
> > copies of an sqlite3 handle which both point to (and own!) the underying
> > memory.
> >
> > NEVER use memcpy() to copy non-trivial types (e.g. those which contain
> > owned
> > pointers).
> >
> > --
> > ----- stephan beal
> > http://wanderinghorse.net/home/stephan/
> >
> >
> > Yes and indeed to "Another point:", thanks for all the suggestions!!
> _______________________________________________
> sqlite-users mailing list
> [email protected]
> http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
_______________________________________________
sqlite-users mailing list
[email protected]
http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
