Mailing-List: contact cygwin-developers-help AT cygwin DOT com; run by ezmlm List-Subscribe: List-Archive: List-Post: List-Help: , Sender: cygwin-developers-owner AT cygwin DOT com Delivered-To: mailing list cygwin-developers AT cygwin DOT com X-Authentication-Warning: atacama.four-d.de: mail set sender to using -f Date: Wed, 22 Jan 2003 16:41:09 +0100 (=?ISO-8859-1?Q?Westeurop=E4ische_Normalzeit?=) From: Thomas Pfaff To: cygwin-developers AT cygwin DOT com Subject: Re: ARGH pthreads tests failing? In-Reply-To: <20030122150338.GA4396@redhat.com> Message-ID: X-X-Sender: pfaff AT antarctica DOT intern DOT net MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII On Wed, 22 Jan 2003, Christopher Faylor wrote: > Or, perhaps more importantly, are you saying that you've broken binary > compatibility with previous versions of cygwin? > In http://cygwin.com/ml/cygwin-developers/2003-01/msg00011.html i wrote: > The pthread mutex patch will change the mutex default type from > recursive to errorchecking (all unix systems that i know are either > errorchecking or fast (deadlocking) by default). > Any objections ? Nobody had any. Changing the default type will not break binary compatibilty, it will only be noticed in cases where recursive types were expected, but: I think it is better to revert a change in pthread.h were the PTHREAD_MUTEX_RECURSIVE and _ERRORCHECK defines has been exchanged. I did not realize that this would break some older code. Reverting this will keep older code working that explicitly set the mutex type to recursive. I do not think that it will make sense to keep the default mutex type recursive if cygwin should be compatible to other pthread implementations and this can not be done without changing PTHREAD_MUTEX_DEFAULT. Thomas