Mailing-List: contact cygwin-help@cygwin.com; run by ezmlm List-Subscribe: List-Archive: List-Post: List-Help: , Sender: cygwin-owner@cygwin.com Mail-Followup-To: cygwin@cygwin.com Delivered-To: mailing list cygwin@cygwin.com Date: Wed, 5 May 2004 20:30:16 -0400 (EDT) From: Steve Thompson Reply-To: smt@vgersoft.com To: cygwin@cygwin.com Subject: Re: pthreads and sockets - Cannot register window class error In-Reply-To: Message-ID: References: <4098760F.CDB4F630@hot.pl> <4098E04B.34D2866B@dessent.net> <20040505210340.GB321@coe.bosbc.com> MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII On Wed, 5 May 2004, Brian Ford wrote: > On Wed, 5 May 2004, Steve Thompson wrote: > > > On Wed, 5 May 2004, Christopher Faylor wrote: > > > > > Unless HOST is a numeric IP address, gethostbyname should be properly > > > reentrant. The only time gethostbyname is not thread safe is when it is > > > resolving a numeric IP. > > > > I believe that gethostbyname(), since it is returning a pointer to a > > statically allocated structure, cannot be thread safe under any > > circumstances. I always wrap it with a mutex until I'm done with the > > hostent structure. > > That's a good portable practice. But, if your going to essentially call > cgf a lyer, please at least check the source. He uses a thread specific > buffer to return that data. It's obvious if you look. Heaven forbid that I should do that; it certainly wasn't my meaning. Apologies to cgf if it was taken that way. In any event, relying on a feature that is particular to one implementation will always bite you in the end. Steve -- Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple Problem reports: http://cygwin.com/problems.html Documentation: http://cygwin.com/docs.html FAQ: http://cygwin.com/faq/