Mailing-List: contact cygwin-developers-help AT cygwin DOT com; run by ezmlm
List-Subscribe: <mailto:cygwin-developers-subscribe AT cygwin DOT com>
List-Archive: <http://sources.redhat.com/ml/cygwin-developers/>
List-Post: <mailto:cygwin-developers AT cygwin DOT com>
List-Help: <mailto:cygwin-developers-help AT cygwin DOT com>, <http://sources.redhat.com/ml/#faqs>
Sender: cygwin-developers-owner AT cygwin DOT com
Delivered-To: mailing list cygwin-developers AT cygwin DOT com
Subject: Re: Many pthread failures in the test suite, one setgroup failure
From: Robert Collins <rbcollins AT cygwin DOT com>
To: cygwin-developers AT cygwin DOT com
In-Reply-To: <20020929000215.GB10872@redhat.com>
References: <20020925141653 DOT GA6134 AT redhat DOT com>
	<1033139976 DOT 22908 DOT 333 DOT camel AT lifelesswks>
	<163544913434 DOT 20020927192540 AT logos-m DOT ru>
	<1033140780 DOT 9593 DOT 0 DOT camel AT lifelesswks>
	<44642850720 DOT 20020928223759 AT logos-m DOT ru>
	<1033254454 DOT 4375 DOT 48 DOT camel AT lifelesswks>  <20020929000215 DOT GB10872 AT redhat DOT com>
Content-Type: multipart/signed; micalg=pgp-sha1; protocol="application/pgp-signature";
	boundary="=-xkv7MUPBU5yiOcGyk0Jb"
Date: 29 Sep 2002 10:50:37 +1000
Message-Id: <1033260638.4375.76.camel@lifelesswks>
Mime-Version: 1.0


--=-xkv7MUPBU5yiOcGyk0Jb
Content-Type: text/plain
Content-Transfer-Encoding: quoted-printable

On Sun, 2002-09-29 at 10:02, Christopher Faylor wrote:

> >Which is valid for verifyable_object but will *kill* any derived
> >classes. Yep. coding bug on my behalf. Can you or Chris try the test run
> >after changing the verifyable_object destructor to be virtual (thread.h,
> >line 163).
>=20
> Already did that.  It brings the pthread test suite failures down to
> 1 for pthread-condvar6.  assertion output is below.

Ok, looking into this in shortly.
=20
> >What *may* be a regression in 3.2 is the apparent bug caused by the
> >introduction of a VMT in a derived class. However, as we already have a
> >destructor for verifyable_object, and in this instance it should be
> >virtual, we get to sidestep the bullet - for now.
>=20
> I was thinking that it made sense to make the destructor virtual anyway.
> But now, I'm worried about other parts of cygwin which are not exercised
> by the test suite.  Maybe they will have problems, too.
>=20
> Isn't this actually a bug?

Uhm what in?
The missing virtual on the destructor *is* a bug in my code.
gcc 3.2 having trouble with derived classes that have virtual methods of
base classes with no virtuals methods is also (IMO) a bug and one in
gcc.

However, there is a thing called the rule of three:
if a class has explictly decalared any of:
copy constructor
assignment operator
destructor

it MUST declare all three. Or else bad things happen (usually with
remote storage). This isn't bad-because-of-compiler, but
bad-because-of-programming-logic-errors.

In this specific case, gcc was doing the right thing. I believe I could
construct a test case to make gcc to the wrong thing though.

Something like:

class myBase{
public:
  myBase (int anInt):_value(anInt){}
  int getvalue() {return _value;}
private:
  int _value;
};

class myDerived {
   myDerived (int anInt) : myBase (anInt);
  virtual void confusebase(){return getvalue();}
  };

should do it.

Rob
--=20
---
GPG key available at: http://users.bigpond.net.au/robertc/keys.txt.
---

--=-xkv7MUPBU5yiOcGyk0Jb
Content-Type: application/pgp-signature; name=signature.asc
Content-Description: This is a digitally signed message part

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.7 (GNU/Linux)

iD8DBQA9lk5dI5+kQ8LJcoIRAtPQAKCPygtlwo6RuXRKIBS0i+ibbJ3wSQCbB5AJ
G1V5hzy1x3F9oBbshTziGTo=
=Vg9L
-----END PGP SIGNATURE-----

--=-xkv7MUPBU5yiOcGyk0Jb--