Mailing-List: contact cygwin-developers-help AT sourceware DOT cygnus DOT com; run by ezmlm
List-Subscribe: <mailto:cygwin-developers-subscribe AT sources DOT redhat DOT com>
List-Archive: <http://sources.redhat.com/ml/cygwin-developers/>
List-Post: <mailto:cygwin-developers AT sources DOT redhat DOT com>
List-Help: <mailto:cygwin-developers-help AT sources DOT redhat DOT com>, <http://sources.redhat.com/ml/#faqs>
Sender: cygwin-developers-owner AT sources DOT redhat DOT com
Delivered-To: mailing list cygwin-developers AT sources DOT redhat DOT com
Message-ID: <3B65A2B8.90702@ece.gatech.edu>
Date: Mon, 30 Jul 2001 14:08:56 -0400
From: Charles Wilson <cwilson AT ece DOT gatech DOT edu>
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.2) Gecko/20010713
X-Accept-Language: en-us
MIME-Version: 1.0
To: cygwin-developers AT sources DOT redhat DOT com
Subject: Re: Problems with autoconf-2.52 testsuite using current CVS Cygwin
References: <996329431 DOT 27668 DOT ezmlm AT sources DOT redhat DOT com> <3B63432E DOT 6050309 AT ece DOT gatech DOT edu> <3B649305 DOT 2090302 AT ece DOT gatech DOT edu> <3B64C0A9 DOT 1080700 AT ece DOT gatech DOT edu> <3B64F567 DOT 6060304 AT ece DOT gatech DOT edu> <3B65835C DOT 9000001 AT ece DOT gatech DOT edu>
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit

Charles Wilson wrote:


> Anyway, after I finish cleaning up ACL's, I'll try the autoconf tests 
> again -- maybe I too will then enjoy the VFORK bug...  :-P


Okay, using a cygwin dll compiled from CVS last night, the mktemp from 
courtesan.com, and running the autoconf tests after fixing all the ACL's 
in sight, I can report:

a) 151 tests passed
b) 4 tests skipped
c) LOTS of empty, unremoved tmp directories in /tmp

I *think* this means I'm now reproducing what everybody else is seeing; 
vfork "bug" and all.  (However, since the directories are not being 
removed, YET I pass all the tests -- I guess courtesan's mktemp 
implementation is clever enough to always generate unique filenames so 
that collision with previously-created-but-not-removed dirnames does not 
cause a test failure).

--Chuck

P.S. Do we need a tool that can

remove the "propagate perms to children" (or "inherit perms from 
parent") property from directories?

remove "extra" ace's from files and dirs -- whatever they happen to be, 
leaving only user:: group:: other:: mask:: (and possibly default:: 
versions of the above, for directories).  That is, remove any user:xxx: 
and group:xxx: specific ACE's?

or can setfacl be used more cleverly than I have been using it(*), to 
automatically do the preceeding?  ((*) I've been explicitly removing the 
"extra" ACE's for user:500: and group:544: since I *know* those are the 
only extra ACE's in my build tree.  I have not been explicitly adding an 
"extra" ACE for group:18: (SYSTEM) which is automatically added for new 
directories under the CVS-cygwin.)

I fear this problem may hit more folks than just me, when 1.3.3 is 
released and Corinna's change from 7/9/2001 takes effect.  Her change 
will prevent *future* problems, but as I demonstrated, can cause 
problems under dirs created using earlier kernels...

OTOH, it only affects W2K/NTFS users...