Date: Fri, 23 Apr 2004 06:40:54 +0400 From: Sergey Zaharchenko <doublef@tele-kom.ru> To: Chris Ashlee <chrisa@uvic.ca> Cc: freebsd-chat@freebsd.org Subject: Re: Beginning C++ in FreeBSD Message-ID: <20040423024054.GA350@Shark.localdomain> In-Reply-To: <40883C6F.5090602@uvic.ca> References: <200404202124.50967.dgw@liwest.at> <FGEIJLCPFDNMGDOKNBABCEAICKAA.flowers@users.sourceforge.net> <20040421110548.20d8e75c.cpressey@catseye.mine.nu> <6.0.1.1.1.20040421191223.03ed1a88@imap.sfu.ca> <p06002031bcac75b028c1@[10.0.1.5]> <20040421124817.5811bddb.cpressey@catseye.mine.nu> <p06002033bcac8db7ca76@[10.0.1.5]> <20040422030636.GA444@Shark.localdomain> <40883C6F.5090602@uvic.ca>
next in thread | previous in thread | raw e-mail | index | archive | help
--C7zPtVaVf+AK4Oqc
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable
On Thu, Apr 22, 2004 at 02:43:11PM -0700,
Chris Ashlee probably wrote:
> DoubleF wrote:
> >[etc] If you think C++ can be translated to C, just translate this:
> >
> >// one module
> >
> >class A
> >{
> >public:
> > A();
> >};
> >
> >A::A()
> >{
> > /* some code here */
> >}
> >
> >static A a;
> >
> >// another module, which doesn't know about the former's existance
> >
> >int
> >main(void)
> >{
> > /* some other code here */
> >}
> >
>=20
> Based on my (limited) knowledge of C++, the constructor for the object=20
> 'a' of type A will be called before main...
Yes, that's the catch...
> So if you were to translate=20
> this to C, the other module, which doesn't know about the A module,=20
> would have to have a call to the constructor inserted before any other=20
> code in main, and 'a' would have to be referenced from the module=20
> containing main.
>=20
> So it seems that some information would be lost in the translation to C,=
=20
> namely that some details of other modules should not be visible.=20
> However, the compiled code for both the C++ program and the C=20
> translation could well wind up the same.
You would have to translate a program as a whole, not on a file-per-ile
basis, then. I agree that in the end you will get something compilable
--- but it will have large dependency chains which aren't really
necessary, and you'd have to recompile the main module each time the A
module's interface changes, which is certainly inefficient...=20
> Any high-level optimizations=20
> that can be done on C++ but not C could probably be done in the=20
> translation itself. Remember, C is basically a glorified, portable=20
> assembly language, and just about anything can be translated to it -=20
> even machine code. It's translating to a more high-level language that's=
=20
> hard.
What you surely can't translate is the A module without the main module.
Suppose I want a shlib to initialize itself when it's loaded. I create a
static object and GCC does the rest of the stuff for me (like creating
an init function visible to the linker). But what would you do in C (of
course, you want it portable, so no init hacks, please...).
--=20
DoubleF
Fourth Law of Revision:
It is usually impractical to worry beforehand about
interferences -- if you have none, someone will make one for you.
--C7zPtVaVf+AK4Oqc
Content-Type: application/pgp-signature
Content-Disposition: inline
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.4 (FreeBSD)
iD8DBQFAiII1wo7hT/9lVdwRApC9AJ4gNgUwH9pIFhJOizQY6vZZgpHBhwCfUAkB
deWeN5cAG9fGlq2nUSBJEII=
=0Ull
-----END PGP SIGNATURE-----
--C7zPtVaVf+AK4Oqc--
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20040423024054.GA350>