From: UnixOS2 Archive To: "UnixOS2 Archive" Date: Tue, 11 Jun 2002 04:27:36 EST-10EDT,10,-1,0,7200,3,-1,0,7200,3600 Subject: [UnixOS2_Archive] No. 239 ************************************************** Monday 10 June 2002 Number 239 ************************************************** Subjects for today 1 Re: c++ and c : John Poltorak 2 Re: Everything is broken! : John Poltorak 3 Re: c++ and c : Holger Veit 4 Re: Everything is broken! : Ted Sikora 5 Re: c++ and c : Andrew Zabolotny" 6 Re: Everything is broken! : Andrew Zabolotny" 7 Re: Everything is broken! : Ted Sikora 8 library numbers : Ted Sikora 9 Re: Everything is broken! : Andreas Buening **= Email 1 ==========================** Date: Tue, 11 Jun 2002 10:06:47 +0100 From: John Poltorak Subject: Re: c++ and c On Tue, Jun 11, 2002 at 12:25:44PM +0400, Andrew Zabolotny wrote: > >Try older versions of gcc, like the original EMX0.9d ( gcc 2.8.1 ) or > >pgcc 2.95; they may be more friendly to the OMF (.obj) format. > I don't see a single reason to use older versions of gcc. Everything that can > be done with older versions can be done with the newer versions. A previous post pointed to this msg on the gcc mailing list:- http://gcc.gnu.org/ml/gcc/2002-04/msg01168.html It has also been mentioned that you get a better Perl build with pgcc 2.95 than gcc 3.0.3 so I can see some reluctance to upgrade at this point. Besides that, I never managed to get Perl built using gcc 3.0.3. It may be my installation, but I have tried many times without any success so far. > > Greetings, > _\ndy > -- John **= Email 2 ==========================** Date: Tue, 11 Jun 2002 10:22:06 +0100 From: John Poltorak Subject: Re: Everything is broken! On Tue, Jun 11, 2002 at 12:27:22PM +0400, Andrew Zabolotny wrote: > On Mon, 10 Jun 2002 19:32:01 -0400, Ted Sikora wrote: > > >Beat you to it.... got it working. We *have* to standardize everything. > >What a mess! No wonder everyone runs from OS/2. > >Anyone with sense or no ambition. > Exactly. When I've rised this question about half of year ago, nobody listened. > I believe nobody will do it this time as well. Erm... I've been trying to get everything standardised for a long time. That was one of the main aims of this list. It is hard going, but I think we have made some progress. We have had quite a lot of updated programs including your gcc updates which are great to have, but one of the problems is that every developer still insists on using his own toolset which often results in no one else being able to rebuild a particular app, or even if it is rebuild it is not possible to provide precise instructions because there may be some depedency on an obsure version of make or a very specific shell which needs to be used. > > Greetings, > _\ndy > > -- John **= Email 3 ==========================** Date: Tue, 11 Jun 2002 11:13:30 +0200 From: Holger Veit Subject: Re: c++ and c On Tue, Jun 11, 2002 at 12:25:44PM +0400, Andrew Zabolotny wrote: > On Mon, 10 Jun 2002 13:35:25 +0100, csaba.raduly at sophos.com wrote: > > >>I'm trying to link c and c++ modules (using ilink): > >Then you should probably use icc, not gcc. > hmm... why? just to link a .c and a .cpp file? ACK. However, you won't take advantage of ilink (incremental linking) with gcc, but it works as well as link386 with gcc. [...] > >Try older versions of gcc, like the original EMX0.9d ( gcc 2.8.1 ) or > >pgcc 2.95; they may be more friendly to the OMF (.obj) format. > I don't see a single reason to use older versions of gcc. Everything that can > be done with older versions can be done with the newer versions. We should negotiate and define what compiler version should be mandatory for UnixOS/2. The problem is the different C++ output 2.8.1, 2.95.x, and 3.0.x produce. This has an important impact on C++ DLLs which occur for instance in KDE. Holger -- Please update your tables to my new e-mail address: holger.veit$ais.fhg.de (replace the '$' with ' at ' -- spam-protection) **= Email 4 ==========================** Date: Tue, 11 Jun 2002 12:23:26 -0400 From: Ted Sikora Subject: Re: Everything is broken! John Poltorak wrote: > > On Tue, Jun 11, 2002 at 12:27:22PM +0400, Andrew Zabolotny wrote: > > On Mon, 10 Jun 2002 19:32:01 -0400, Ted Sikora wrote: > > > > >Beat you to it.... got it working. We *have* to standardize everything. > > >What a mess! No wonder everyone runs from OS/2. > > >Anyone with sense or no ambition. > > Exactly. When I've rised this question about half of year ago, nobody listened. > > I believe nobody will do it this time as well. > > Erm... I've been trying to get everything standardised for a long time. > That was one of the main aims of this list. It is hard going, but I think > we have made some progress. We have had quite a lot of updated programs > including your gcc updates which are great to have, but one of the > problems is that every developer still insists on using his own toolset > which often results in no one else being able to rebuild a particular app, > or even if it is rebuild it is not possible to provide precise > instructions because there may be some depedency on an obsure version of > make or a very specific shell which needs to be used. > I was just blowing off steam. That's what UnixOS2/Posix is doing! Just gets frustrating messing with all the 'old' stuff and mixing it. I think the dev/build environment is priority. That way what's ported today will still run years from now. Now that's a pipe dream. More likely on OS/2. Forget Linux if it's 3 months old you might as well trash it.(binaries) -- Ted Sikora tsikora at unixos2.com http://unixos2.com **= Email 5 ==========================** Date: Tue, 11 Jun 2002 12:25:44 +0400 From: "Andrew Zabolotny" Subject: Re: c++ and c On Mon, 10 Jun 2002 13:35:25 +0100, csaba.raduly at sophos.com wrote: >>I'm trying to link c and c++ modules (using ilink): >Then you should probably use icc, not gcc. hmm... why? just to link a .c and a .cpp file? >>The output from my build: >>test > gmake -f gcchello.mak USE_ILINK=1 test.exe >>gcc -c -Zsys -Zmt -Zomf test.c -otest.obj >>g++ -c -Zsys -Zmt -Zomf x.cpp -ox.obj >>ilink /NOFREE test.obj x.obj d:\emx\lib\crt0.obj /NOD, test.exe, test.map, >>gcc.lib omflib.lib c_alias.lib end.lib d:\emx\lib\mt\c.lib >>d:\emx\lib\mt\c_app.lib d:\emx\lib\mt\sys.lib os2386.lib; >> >>IBM(R) Linker for OS/2(R), Version 03.06.PPK1981104 >>Copyright (C) IBM Corporation 1988, 1998. >>Copyright (C) Microsoft Corp. 1988, 1989. >>All rights reserved. >> >> >>x.obj(x.obj) : error LNK2029: "__gxx_personality_v0" : unresolved external >>x.obj(x.obj) : error LNK2029: "_ZN4Test4testEv" : unresolved external __gxx_personality_v0 is defined in gcc.lib. You forgot to add it to the library list. This is a dummy symbol which is referenced from any g++-produced object file (not sure why, it has something to do with the version of de/mangler and such). >gcc 3 introduces all kinds of advanced C++ features. Their implementation >is sometimes incompatible with the OMF format. That's not quite correct. All the features are implemented in OMF, but sometimes you will need additional manipulations, which are described in details in the docs. >Try older versions of gcc, like the original EMX0.9d ( gcc 2.8.1 ) or >pgcc 2.95; they may be more friendly to the OMF (.obj) format. I don't see a single reason to use older versions of gcc. Everything that can be done with older versions can be done with the newer versions. Greetings, _\ndy **= Email 6 ==========================** Date: Tue, 11 Jun 2002 12:27:22 +0400 From: "Andrew Zabolotny" Subject: Re: Everything is broken! On Mon, 10 Jun 2002 19:32:01 -0400, Ted Sikora wrote: >Beat you to it.... got it working. We *have* to standardize everything. >What a mess! No wonder everyone runs from OS/2. >Anyone with sense or no ambition. Exactly. When I've rised this question about half of year ago, nobody listened. I believe nobody will do it this time as well. Greetings, _\ndy **= Email 7 ==========================** Date: Tue, 11 Jun 2002 18:20:33 -0400 From: Ted Sikora Subject: Re: Everything is broken! Andreas Buening wrote: > > John Poltorak wrote: > > > > On Tue, Jun 11, 2002 at 12:27:22PM +0400, Andrew Zabolotny wrote: > > > On Mon, 10 Jun 2002 19:32:01 -0400, Ted Sikora wrote: > > > > > > >Beat you to it.... got it working. We *have* to standardize everything. > > > >What a mess! No wonder everyone runs from OS/2. > > > >Anyone with sense or no ambition. > > > Exactly. When I've rised this question about half of year ago, nobody listened. > > > I believe nobody will do it this time as well. > > > > Erm... I've been trying to get everything standardised for a long time. > > That was one of the main aims of this list. It is hard going, but I think > > we have made some progress. We have had quite a lot of updated programs > > including your gcc updates which are great to have, but one of the > > problems is that every developer still insists on using his own toolset > > which often results in no one else being able to rebuild a particular app, > > or even if it is rebuild it is not possible to provide precise > > instructions because there may be some depedency on an obsure version of > > make or a very specific shell which needs to be used. > > You seem to be the main package distributor. Then _define_ > how packages have to be build before they can be included > into UO/2. You will never get an agreement on this list > (or whereever) which tools to use. Consider a solution > that is as simple as possible and as complex as necessary > to maximize flexibility and minimize effort for maintaining > and then make a decision. Put it onto a web page so that > everybody can read it every day without producing endless > threads. Either enough people will follow you or there > will never be a real organized UO/2 release. > > bye, > Andreas > John could you come up with a guideline? I would think everything should be ported using UnixOS/2's dependancies as a guideline. Seems everyone has a favorite library. If you use the ports from os2.ru you need their dll's. Zlib, tiff, gtk/glib, and png are good examples for gui apps. There's HU's, UnixOS2, Warpzilla, etc. Graphic libs and what development environment like EMX 0.9D or PGCC with 2.95 dll's should be recommended. I think EMX runtimes should be for old ports and legacy apps only. Everything now should depend on 2.95.3 or newer. We should draw the line now. We need a standard now that everyone, Warpzilla etc. follows. Maybe getting together with Henry is a good idea too. -- Ted Sikora tsikora at unixos2.com http://unixos2.com **= Email 8 ==========================** Date: Tue, 11 Jun 2002 19:00:52 -0400 From: Ted Sikora Subject: library numbers We should implement a numbering convention either like the Linux libz.so.1.1.4 or FreeBSD aka libz.so.4 That way we can still use multiple libraries and not resort to patchwork like renaming them. -- Ted Sikora tsikora at unixos2.com http://unixos2.com **= Email 9 ==========================** Date: Tue, 11 Jun 2002 22:56:29 +0200 From: Andreas Buening Subject: Re: Everything is broken! John Poltorak wrote: > > On Tue, Jun 11, 2002 at 12:27:22PM +0400, Andrew Zabolotny wrote: > > On Mon, 10 Jun 2002 19:32:01 -0400, Ted Sikora wrote: > > > > >Beat you to it.... got it working. We *have* to standardize everything. > > >What a mess! No wonder everyone runs from OS/2. > > >Anyone with sense or no ambition. > > Exactly. When I've rised this question about half of year ago, nobody listened. > > I believe nobody will do it this time as well. > > Erm... I've been trying to get everything standardised for a long time. > That was one of the main aims of this list. It is hard going, but I think > we have made some progress. We have had quite a lot of updated programs > including your gcc updates which are great to have, but one of the > problems is that every developer still insists on using his own toolset > which often results in no one else being able to rebuild a particular app, > or even if it is rebuild it is not possible to provide precise > instructions because there may be some depedency on an obsure version of > make or a very specific shell which needs to be used. You seem to be the main package distributor. Then _define_ how packages have to be build before they can be included into UO/2. You will never get an agreement on this list (or whereever) which tools to use. Consider a solution that is as simple as possible and as complex as necessary to maximize flexibility and minimize effort for maintaining and then make a decision. Put it onto a web page so that everybody can read it every day without producing endless threads. Either enough people will follow you or there will never be a real organized UO/2 release. bye, Andreas -- One OS to rule them all, One OS to find them, One OS to bring them all and in the darkness bind them In the Land of Redmond where the Shadows lie.