Received: from PACIFIC-CARRIER-ANNEX.MIT.EDU by po7.MIT.EDU (5.61/4.7) id AA18770; Thu, 22 Feb 96 22:08:30 EST
Received: by MIT.EDU 
	id AB07322; Thu, 22 Feb 96 22:06:49 EST
Date: Thu, 22 Feb 96 22:06:49 EST
From: MAILER-DAEMON@MIT.EDU (Mail Delivery Subsystem)
Subject: Returned mail: Cannot send message for 3 days
Message-Id: <9602230306.AB07322@MIT.EDU>
To: <proven@MIT.EDU>
To: proven@PO7.LOCAL

   ----- Transcript of session follows -----
421 clem.systemsix.com.tcp... Deferred: Connection timed out during user open with clem.systemsix.com
451 clem@clem.systemsix.com... Cannot send message for 3 days
clem@clem.systemsix.com... Cannot send message for 3 days

   ----- Unsent message follows -----
Received: from YAZ-PISTACHIO.MIT.EDU by MIT.EDU with SMTP
	id AA24974; Mon, 19 Feb 96 20:22:34 EST
Received: by yaz-pistachio.MIT.EDU (5.57/4.7) id AA04308; Mon, 19 Feb 96 20:22:54 -0500
Message-Id: <9602200122.AA04308@yaz-pistachio.MIT.EDU>
To: "Marty Leisner" <leisner@sdsp.mc.xerox.com>
Cc: pthreads@MIT.EDU
Subject: Re: Anybody used Linux libc support for pthreads? 
In-Reply-To: Your message of "Mon, 19 Feb 1996 11:10:41 PST."
             <9602191910.AA14798@gnu.mc.xerox.com> 
Date: Mon, 19 Feb 1996 20:22:52 EST
From: Christopher Provenzano  <proven@MIT.EDU>


The last time I communicated with HJ he was planning on making the Linux
C library thread safe and take and distributing a release of pthreads
with it. Eventually I would make a compile option for pthreads to only
compile the thread routines that would be a drop in replacement for the
pthread library supplied. This way the Linux C library and pthreads 
could continue development separately. (Similar work is also underway 
for NetBSD and FreeBSD.)

My distribution of pthreads will continue to ship a different variant of
a thread safe C library based on the BSD C library. There are some 
differences between the two C libraries so you should pick one and stick 
with it and know what the differences are. The biggest one I know of is
how we deal with routines that have static data. I supply routines
that use thread specific data and new _r variants whereas HJ only supplies 
the _r variants.

CAP
