Subject: Re: fork(2) vs. pthread_create() (fwd)
To: Nathan J. Williams <nathanw@wasabisystems.com>
From: Love <lha@stacken.kth.se>
List: tech-userlevel
Date: 06/10/2004 20:50:34
--=-=-=


"Nathan J. Williams" <nathanw@wasabisystems.com> writes:

> manu@NetBSD.org (Emmanuel Dreyfus) writes:
>
>> I did that already. But as this works on other systems, and even on
>> NetBSD with GNU Pth, I'm sure someone else will get bitten by this
>> problem later.
>
> There are lots of areas of standards where the behavior of some
> operation or set of operations is undefined. I don't think that we're
> generally obligated to define and fix undefined things that other OS's
> happen to have; and as much as I like the option, I also don't think
> it's worthwhile to make all undefined things fail obviously, to
> prevent programmers from accidentally using them.

But this time its quite simple to make sure the timer is (re)created since
you'll know when it no longer there (see my patch).

I don't mind not fixing the problem. there is enough problems with trying
to use (p)threads that applications/libraries shouldn't do any fancy stuff.

Love


--=-=-=
Content-Type: application/pgp-signature

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.4 (NetBSD)

iQEVAwUAQMitfHW+NPVfDpmCAQJPAgf/YOTjH7fLrod+eSRiAfg5CzZ8scnW6uMY
vNrKmIAl81SF62KfjFWzKSGN2k7wodYwpVtk48m6/t19SPiPIyEt8fyQA2NS0O69
oYqjcvyGczqtEXUfbOeJQPW6C6H0UgSmS/LRVfYDImN2C9q5ZrxTjhHq8N3GEb5l
yt1fhoxoUHgTyc8uLIFgpn1mgz7+HbKf/mulD96DrJTwu8xgqaTMBl41bU5XEY2L
Egkq449bBi+UhHHgKtR/Pdd6D9K59rgqQEYFmGarQaX5bnYP1MyebFv3saioKgEh
2fStTz3YILaIC/Q+Qngufq+4IC4RZybAq38WYjUv3gVj/jKu5QQHvg==
=iWVT
-----END PGP SIGNATURE-----
--=-=-=--