Mailing-List: contact cygwin-help AT cygwin DOT com; run by ezmlm List-Subscribe: List-Archive: List-Post: List-Help: , Sender: cygwin-owner AT cygwin DOT com Mail-Followup-To: cygwin AT cygwin DOT com Delivered-To: mailing list cygwin AT cygwin DOT com Date: Mon, 30 Sep 2002 08:51:11 -0400 From: Jason Tishler Subject: Fwd: Fw: [Mingw-users] Whats the latest on libtool and the stub dll libraries? To: Norman Vine Cc: Cygwin Mail-followup-to: Norman Vine , Cygwin Message-id: <20020930125111.GA1580@tishler.net> MIME-version: 1.0 Content-type: multipart/mixed; boundary="Boundary_(ID_7bkorFsXIXTaM5wphqSQSA)" User-Agent: Mutt/1.4i Note-from-DJ: This may be spam --Boundary_(ID_7bkorFsXIXTaM5wphqSQSA) Content-type: text/plain; charset=us-ascii Content-transfer-encoding: 7BIT Content-disposition: inline Norman, Please post instead of sending private email. On Fri, Sep 27, 2002 at 10:53:26PM -0400, Norman Vine wrote: > Is it the -mdll flag that is causing our dll relocation problems > with the extension modules that are built with setup.py ??? No. IMO, the above is not relevant to the Cygwin fork()/rebase issue. Note that the "-mdll" flag is used when building Win32 Python shared extension modules with Cygwin gcc in "-mno-cygwin" mode -- not when building Cygwin Python shared extensions. BTW, Cygwin fork() has rebase issues with Apache, Perl, etc. too. On Sat, Sep 28, 2002 at 12:14:49AM -0400, Norman Vine wrote: > and the following (3) modules had the remap error for cygssl.dll > when run from inside the test-suite > test_fork1.py > test_popen2.py > test_pty.py > but ran successfully when run directly The above is the expected behavior. When the entire Python regression test is run, the probability of a DLL base address conflict is much higher than when individual regression tests are run separately. Jason --Boundary_(ID_7bkorFsXIXTaM5wphqSQSA) Content-type: message/rfc822 Received: from cust_req_fwding (jason AT tishler DOT net --> jt AT tishler DOT net) by lmg.ahnet.net id <291908-752>; Fri, 27 Sep 2002 21:14:39 -0700 Received: from mx5.cape.com ([204.107.252.38]) by lmg.ahnet.net with ESMTP id <292904-755>; Fri, 27 Sep 2002 21:14:32 -0700 Received: from sfdev3 (tsc-91.cape.com [140.186.55.91]) by mx5.cape.com (8.12.3/8.12.3) with SMTP id g8S4ES1I002503 for ; Sat, 28 Sep 2002 00:14:29 -0400 Date: Sat, 28 Sep 2002 00:14:49 -0400 From: Norman Vine Subject: Fw: [Mingw-users] Whats the latest on libtool and the stub dll libraries? To: Jason Tishler Reply-to: Norman Vine Message-id: <000e01c266a5$97164010$0100a8c0 AT sfdev3> MIME-version: 1.0 X-MIMEOLE: Produced By Microsoft MimeOLE V6.00.2600.0000 X-Mailer: Microsoft Outlook Express 6.00.2600.0000 Content-type: text/plain; charset=iso-8859-1 Content-transfer-encoding: 7BIT X-Priority: 3 X-MSMail-priority: Normal X-Virus-Scanned: Cape.Com VirusScan, no known virus found X-Spam-Status: No, hits=0.5 required=5.0 tests=SUBJ_ENDS_IN_Q_MARK,PORN_10 version=2.31 X-Spam-Level: Hi Jason I just did a quick test compilng a ~month old CVS version of Python after making the following change in distutills.cygwincompiler # Hard-code GCC because that's what this is all about. # XXX optimization, warnings etc. should be customizable. self.set_executables(compiler='gcc -mcygwin -O -Wall', # compiler_so='gcc -mcygwin -mdll -O -Wall', compiler_so='gcc -mcygwin -shared -O -Wall', linker_exe='gcc -mcygwin', # linker_so=('%s -mcygwin -mdll -static' % linker_so=('%s -mcygwin -shared' % self.linker_dll)) and the following (3) modules had the remap error for cygssl.dll when run from inside the test-suite test_fork1.py test_popen2.py test_pty.py but ran successfully when run directly HTH Norman < I actually changed all occurences of gcc in cygwincompiler to gcc-2 in that I am using the 'beta' compiler > ----- Original Message ----- From: "Norman Vine" To: "Jason Tishler" Sent: Friday, September 27, 2002 10:53 PM Subject: Fw: [Mingw-users] Whats the latest on libtool and the stub dll libraries? > Jason > > FYI > > from distutils.cygwinccompiler ~line 100 > > self.set_executables(compiler='gcc -mcygwin -O -Wall', > compiler_so='gcc -mcygwin -mdll -O -Wall', > linker_exe='gcc -mcygwin', > linker_so=('%s -mcygwin -mdll -static' % > self.linker_dll)) > > Is it the -mdll flag that is causing our dll relocation problems > with the extension modules that are built with setup.py ??? > > Norman > > ----- Original Message ----- > From: "Danny Smith" > To: "Elizabeth Barham" > Cc: > Sent: Friday, September 27, 2002 10:39 PM > Subject: Re: [Mingw-users] Whats the latest on libtool and the stub dll > libraries? > > > > > > ----- Original Message ----- > > From: "Earnie Boyd" > > To: "Elizabeth Barham" > > Cc: > > Sent: Saturday, 28 September 2002 01:31 > > Subject: Re: [Mingw-users] Whats the latest on libtool and the stub dll > > libraries? > > > > > > > Elizabeth Barham wrote: > > > > > > > > Earnie Boyd writes: > > > > > > > > > I found the reason, and I must say Elizabeth did a nice job. > > > > > > > > Thanks, but.... it did not work on my copy of MSYS unfortunately. > > When > > > > building the dll, the -mdll was added onto the command line and gcc > > > > errored out saying they were incompatible. Oh well. > > > > > > > Hmm... I didn't get that at all. Make sure you remove > > > /etc/config.site. As for -mdll, it may be useful on the object build. > > > Danny Smith, can you speak to the -mdll switch effect in cc1? The > > > --help gives, "Generate code for a DLL". > > > > > > > -mdll doesn't set any internal flags for cc1.exe, nor does -shared. > > > > -mdll tells driver to pass dllcrt2.o rather than crt2.o as startup > > object to ld > > and to use _DllMainCRTStartup AT 12 as entry point address. -shared does > > this too. > > > > > > -mdll passes --dll to ld. -shared passes -shared to ld. This is where > > the difference is: > > > > --dll only says: 'use default image base for dlls'. It doesn't do any > > of the relocateability magic. > > -shared says: "build a relocateable dll (or more generally, a shareable > > library)" ie, do all the things that dllwrap/dlltool would do > > > > HTH. > > Danny > > > > > > > > > > ------------------------------------------------------- > > This sf.net email is sponsored by:ThinkGeek > > Welcome to geek heaven. > > http://thinkgeek.com/sf > > _______________________________________________ > > MinGW-users mailing list > > MinGW-users AT lists DOT sourceforge DOT net > > > > You may change your MinGW Account Options or unsubscribe at: > > https://lists.sourceforge.net/lists/listinfo/mingw-users > > > --Boundary_(ID_7bkorFsXIXTaM5wphqSQSA) Content-Type: text/plain; charset=us-ascii -- Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple Bug reporting: http://cygwin.com/bugs.html Documentation: http://cygwin.com/docs.html FAQ: http://cygwin.com/faq/ --Boundary_(ID_7bkorFsXIXTaM5wphqSQSA)--