To: cygwin AT cygwin DOT com
From: Mathias Bauer <Mathias DOT Bauer AT oracle DOT com>
Subject: core dump using Make 3.81
Date: Wed, 16 Feb 2011 10:26:17 +0100
Lines: 42
Message-ID: <ijg57q$9fr$1@dough.gmane.org>
Mime-Version: 1.0
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (Windows; U; Windows NT 6.1; en-US; rv:1.9.2.13) Gecko/20101207 Lightning/1.0b2 Mnenhy/0.8.3 OracleBeehiveExtension/1.0.0.2-OracleInternal ObetStats/CATCAFCAFCAFCAFCAFLAF_1292428138647-396660266 Thunderbird/3.1.7
Mailing-List: contact cygwin-help AT cygwin DOT com; run by ezmlm
Sender: cygwin-owner AT cygwin DOT com
Mail-Followup-To: cygwin AT cygwin DOT com

Hi,

I'm working on OpenOffice.org and we are currently reworking our build 
system. Our current combo of dmake and some home made perl scripts shall 
be replaced by GNU Make and a few awk scripts.

We finished our work for building C++/C/Objective-C code and now use 
this on Windows, Linux, Solaris, MacOSX, FreeBSD in some of our modules 
where new make files already have been written. On Windows we have a 
mingw port and a cygwin port.

On all platforms we are using GNU Make 3.81. It works fine in most 
cases, but on cygwin we got some crashes when building in parallel with 
-j. The reason is

http://savannah.gnu.org/bugs/?20033

This bug was fixed already in upstream GNU Make and the fix was 
integrated into the 3.82 release. As we never got one of these crashes 
on Linux, it seems that the Linux distros have integrated this patch 
into their 3.81 releases. An upgrade to 3.82 would fix that for cygwin 
also. As I understood from reading here, there are plans to update the 
cygwin Make package to 3.82.

So far, so good.

But when we tried to use a 3.82 version we built from the official 
source package ourselves, we found a serious performance problem in this 
version. This problem is in the cygwin version only, not in the Linux 
version. As an example, we have a module where checking dependencies 
with 3.81 takes approximately 5 seconds on cygwin (less than 2 seconds 
on Linux BTW), with 3.82 it took 1:30 minutes. So I'm afraid that 
waiting for 3.82 won't be an option for us.

For future plans it would be interesting for us to know if a patch we 
provided to fix that performance issue (so far we don't have this patch) 
would have a chance to get integrated into the official cygwin package. 
If not, we have to build our own make.exe anyway and probably would 
stick with a patched 3.81.

Best regards,
Mathias


--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple