- To: Thomas Bushnell BSG <email@example.com>
- Cc: debian-hurd <firstname.lastname@example.org>
- Subject: Re: pthread_atfork()
- From: Samuel Thibault <email@example.com>
- Date: Fri, 28 Dec 2007 02:04:05 +0000
- Message-id: <20071228020405.GF4314@implementation>
- Mail-followup-to: Thomas Bushnell BSG <firstname.lastname@example.org>, debian-hurd <email@example.com>
- In-reply-to: <1198805101.19780.21.camel@localhost>
- References: <20071222013540.GA4695@implementation> <1198289412.8917.6.camel@localhost> <20071223205303.GE5098@implementation> <1198529117.4724.44.camel@localhost> <20071228011449.GE4314@implementation> <1198805101.19780.21.camel@localhost>
Thomas Bushnell BSG, le Thu 27 Dec 2007 17:25:01 -0800, a écrit :
> > Ah, you mean in <gnu/stubs.h> I guess? However, that's a GNU extension,
> > which pike7.6 will most probably not want to depend on.
> It's using *configure*. That's a GNU thing already. If it's using
> configure, and not wanting to do gnu things, then there is no reason to
> fret about them further.
There is a difference between using GNU configuration tools and
targetting GNU systems in particular. But as you point out:
> Are you aware that every autoconf-using program *already* does this,
Nope. There are so many things like this that in an ideal world
everybody should be aware of...
Actually, it happens that in the pike7.6 case, it doesn't, because its
configure.in doesn't use AC_CHECK_FUNCS for that function. Yes, that's
evil, but that's unfortunately what people do.
Now, that said, we'd have to modify libc's <gnu/stubs.h> to also include
a <gnu/stubs-pthread.h> provided by the hurd's libpthread.