From: UnixOS2 Archive To: "UnixOS2 Archive" Date: Mon, 28 Oct 2002 04:39:40 EST-10EDT,10,-1,0,7200,3,-1,0,7200,3600 Subject: [UnixOS2_Archive] No. 356 ************************************************** Sunday 27 October 2002 Number 356 ************************************************** Subjects for today 1 RE: eComStation 1.1 : Dave Webster 2 Re: Roadmap? : John Poltorak 3 Re: Roadmap? : John Poltorak 4 Re: Roadmap? : Adrian Gschwend" 5 Re: R port? : Thomas Hoffmann 6 Re: Roadmap? : IanM" **= Email 1 ==========================** Date: Mon, 28 Oct 2002 09:57:21 -0600 From: Dave Webster Subject: RE: eComStation 1.1 Will it be availabe from Prism in the US at that time? No announcements at all from any of the Web Sites concerning this. -----Original Message----- From: Christian Hennecke [mailto:christian.hennecke at ruhr-uni-bochum.de] Sent: Friday, October 25, 2002 10:44 AM To: os2-unix at eyup.org Subject: Re: eComStation 1.1 On Fri, 25 Oct 2002 09:44:10 -0500, Dave Webster wrote: >Anybody have any clue when this will finally be ready? Been waiting some >time and the news is pretty sketchy. All I've been able to find is some >screen shots of the new installer. If not soon, I'm just going to stick to >Warp as I now have an inside source for all the latest patches, FP16 and so >on, and drivers. My understanding is that eCS 1.1 will be released at Warpstock Europe, i.e. November 8th. NLVs will follow some weeks later. Christian Hennecke **= Email 2 ==========================** Date: Mon, 28 Oct 2002 11:22:50 +0000 From: John Poltorak Subject: Re: Roadmap? On Sun, Oct 27, 2002 at 09:37:10AM -0600, Jeff Robinson wrote: > Hi folks, > > I was wondering if there is anything resembling a roadmap for getting > UnixOS2 moving forward? > > I know for a while, there was an incredible amount of activity on this > list but to be honest, I'm not certain at what point we sit right now. > Perhaps mostly what I'm looking for is: > > - What would quantify a 1.0 "release" > - target programs to have compile > - the state of programs needed for a 1.0 release > - if they build, don't build, haven't tried, what needs to be done > - documentation > - Andreas has already laid out a filesystem framework with unixos2_fhs.txt > - porting guidelines/logistics > > It would be interesting to see if something like this could be compiled > so it could be made available via unixos2.com. Perhaps a way that folks > could see what others are working on so we don't double up. This may > also help give a more easily quantifiable way of seeing where we are in > the whole scheme of things. > > If something like this doesn't exist yet, I could be a likely candidate > for trying to organise this information into something useful. > > I think that it is not only useful to see for the folks contributing so > that they can see what is going on, but also from a psychological POV as > it is a bit easier to tell that things are indeed moving forward. The whole UnixOS/2 project seems to have stalled because of lack of interest. This is a great pity because a large chunk of a potential distro has been assembled, an FTP repository is in place, and as a result of a considerable amount of discussion over the years a basic structure was emerging. We even have a custom built pkg installer. And a number of key packages have been brought uptodate. After some initial debate many months ago, we had adopted Slackware as Linux distro to model UnixOS/2 on, and I spent around six months putting together OS/2 equivalents of around 90 Slackware packages, using the most uptodate OS/2 copies of any binaries I could find. This was all in place on unixos2.com at one time, but may well have disappeared. Over time it seems to have become apparent that no one knows how to use Slackware pkgs anyway. Maybe people were expecting to be able to download a complete distro and have it up and running on the click of an icon... People forget that the whole UnixOS/2 distro is largely work in progress and has been anywhere near a v1.0 release, as a single package. It also emerged that Slackware was not seem very FHS compliant. Since the assembly of the distro didn't seem to be making any progress, I thought it would be a good idea to try and define a basic toolset which could be used to build packages from source and this did seem to work very well for the purpose of building Perl in a 'clean' environment. As a result of quite a lot of feedback and some excellent contributions, we did manage to come up with a method of building Perl v5.8.0 for OS/2 on the day the source was released. IMV that was a major achievement, and it would be great if we could do that with all Open Source Software without requiring the intervention of the OS/2 maintainer to do his bit. I'd like to see the same situation with the likes of gcc, emacs and all the other big programs that we are fmiliar with. Some months ago, I proposed a UnixOS/2 build system which would incorporate all OS/2 ports of Unix programs, and this would allow users to build binaries for themselves using a standard BUILD program. This would 'empower' users to embed whatever drive letters, if any, in their programs instead of complaining that they were built with 'C:' and their own UnixOS/2 system was on 'F:'. The basis of that Build System was used for building Perl, and I was hoping to extend it, but didn't get much feedback. I think having a standard build system is essential part of making any progress towards a UnixOS/2 distro because as things stand one porter may develop a program which works fine for him but has a prerequisite for a particular 'something' which other people don't have so the program won't work. The porter may not even be aware of of what that 'something' is. It may be an environment variable, a very specific version of a particular DLL, something in the PATH, or maybe a thousand other things, but without a standard build environment it is impossible to guarantee that a program will build or work, on two different machines. A baseline toolset is something which I have suggested as a starting point for the build system, and as was shown by the build of Perl, this can ensure that people can manage to build something quite complex with little or no previous experience of building anything from scratch. The results that several people submitted were extremely similar and it would have been useful to carry on from that point. Maybe I should try an get my proposed Build Sytem uploaded to unixos2.com so that it can be tested more fully... > Jeff > > -- > ---------------- > Whatza JamochaMUD? > http://jamochamud.anecho.mb.ca > > Or other stuff: http://www.anecho.mb.ca/~jeffnik > ----------------------------------------------------------- -- John **= Email 3 ==========================** Date: Mon, 28 Oct 2002 13:49:38 +0000 From: John Poltorak Subject: Re: Roadmap? On Mon, Oct 28, 2002 at 10:25:15PM +1100, IanM wrote: > Hi John > > >After some initial debate many months ago, we had adopted Slackware as > >Linux distro to model UnixOS/2 on, and I spent around six months putting > >together OS/2 equivalents of around 90 Slackware packages, using the most > >uptodate OS/2 copies of any binaries I could find. This was all in place > >on unixos2.com at one time, but may well have disappeared. > > I'm missing some stuff ? > If I am, tell me and I'll see about getting it (back). Since Slackware was agreed as the reference platform, the original directory structure was devised to be as Slackware like as possible, so the different series of packages (a1, ap1, d1, etc) were dumped in:- ftp://unixos2.com/pub/unixos2/unixos2-current/unixos2/ a1 ap1 d1 ... I even got the developer of Less to post a direct link to what I hoped would always the most uptodate version of Less for OS/2, which you can see at:- http://www.greenwoodsoftware.com/less/download.html ie.:- 3. OS/2 Binaries OS/2 binaries are packaged in a zip archive. Binaries for OS/2 are available at http://unix.os2site.com. The latest less binary can be downloaded directly from ftp://unixos2.com/pub/unixos2/unixos2-current/unixos2/a1/less.zip. This did work at one time. I also posted many links to available archives which were correct at the time they were posted. All the archives are probably on your site now, but I don't know if they have been restored. > >Over time it seems to have become apparent that no one knows how to use > >Slackware pkgs anyway. Maybe people were expecting to be able to download > >a complete distro and have it up and running on the click of an icon... > > I think they expect to have there hands held to much sometimes. :-( > But personally I think what you have done is great but.... > > >People forget that the whole UnixOS/2 distro is largely work in > >progress and has been anywhere near a v1.0 release, as a single package. > >It also emerged that Slackware was not seem very FHS compliant. > > The Slackware was a good idea at the time but both myself and > others have found it confusing. You did take leadership though > and it had to start somewere ! Having Slackware as a reference did give us a framework to work with and I still think it could be useful, but I'd like to rebuild all the exist archives from scratch using a specified set of tools and a definitive build script. Once a build script is established, then it can easily accomodate a target package format based on Slackware, or even anything else as a final step in the build process. > >Since the assembly of the distro didn't seem to be making any progress, I > >thought it would be a good idea to try and define a basic toolset which > >could be used to build packages from source and this did seem to work very > >well for the purpose of building Perl in a 'clean' environment. As a > >result of quite a lot of feedback and some excellent contributions, we did > >manage to come up with a method of building Perl v5.8.0 for OS/2 on the > >day the source was released. IMV that was a major achievement, and it > >would be great if we could do that with all Open Source Software without > >requiring the intervention of the OS/2 maintainer to do his bit. I'd like > >to see the same situation with the likes of gcc, emacs and all the other > >big programs that we are fmiliar with. > > This is were we need "maintainers" to submit OS/2 specific changes > to the original code maintainers, and is also part of the stumbling > block, not all original code maintainers are interested in going to > the trouble to add OS/2, or even Windows specific changes, when > they do though, thats great. > > The main problem I've come across over the past few years > in trying to compile packages (and I dont claim to be an expert) > is were the OS/2 specific changes included in the porters code > that has been included with the earlier binaries appears to be > incomplete, missing a diff file, and lack of documentation as to > what had to be done. I can understand the documentation part, > as its not always easy to remember what was done but generating > a diff file against the original code would make life so much easier > (as some do thank goodness) and would be a life saver for some > projects. > > >Some months ago, I proposed a UnixOS/2 build system which would > >incorporate all OS/2 ports of Unix programs, and this would allow users to > >build binaries for themselves using a standard BUILD program. This would > >'empower' users to embed whatever drive letters, if any, in their programs > >instead of complaining that they were built with 'C:' and their own > >UnixOS/2 system was on 'F:'. The basis of that Build System was used for > >building Perl, and I was hoping to extend it, but didn't get much > >feedback. I think having a standard build system is essential part of > > >making any progress towards a UnixOS/2 distro because as things stand one > >porter may develop a program which works fine for him but has a > >prerequisite for a particular 'something' which other people don't have so > >the program won't work. The porter may not even be aware of of what that > > So true but on AIX I've also found this a problem, just not to the same > extent. > > >'something' is. It may be an environment variable, a very specific version > >of a particular DLL, something in the PATH, or maybe a thousand other > >things, but without a standard build environment it is impossible to > >guarantee that a program will build or work, on two different machines. > > So far, I've managed to successfully compile 2 packages out of about > 20 under OS/2, makes me feel like a completely hopeless case, I also > spent many a long night trying to figure out what was wrong with my > setup, only to give up in frustration. Yes, I know the feeling. It's even worse when you successfully build something one day, but when you try it again to make sure, it just falls flat. That is one good thing about so many people being able to build Perl on their own, using the same build script. > >A baseline toolset is something which I have suggested as a starting point > >for the build system, and as was shown by the build of Perl, this can > >ensure that people can manage to build something quite complex with little > >or no previous experience of building anything from scratch. The results > >that several people submitted were extremely similar and it would have > >been useful to carry on from that point. > > >Maybe I should try an get my proposed Build Sytem uploaded to unixos2.com > >so that it can be tested more fully... > > Um, yes please, I was going to query you re your build system :-) Maybe you could incorporate this build system somewhere on unixos2.com... It is still work in progress, but I think it's a good starting point for building apps. > Part of the stalling might also be related to the undated GCC, I havent > make any attempts to recompile anything myself for a while, as I've > been patiently waiting for the latest comilers to be reasonably stable, > or has this happened and I've missed it. I gave up on gcc v3.0.3 because I couldn't get it to build Perl. I would prefer to use something tried and tested such as v2.8.1 until all the basic toolset is complete. > Cheers > IanM > http://www.os2site.com/ > > > Iraqi rifle for sale. Never fired. Dropped once. -- John **= Email 4 ==========================** Date: Mon, 28 Oct 2002 16:56:23 +0100 (CET) From: "Adrian Gschwend" Subject: Re: Roadmap? On Sun, 27 Oct 2002 09:37:10 -0600, Jeff Robinson wrote: Hi Jeff, >I was wondering if there is anything resembling a roadmap for getting >UnixOS2 moving forward? not that I know but a roadmap definitely wouldn't hurt >- What would quantify a 1.0 "release" > - target programs to have compile IMHO we should especially target on development tools to make sure that people can compile other applications more or less easy on OS/2. This means GCC, autoconf & co, perl, make and the base libraries should be finished first. Like this other people will probably start trying to port some other code as well (i.e. myself :-) I simply didn't start yet because I don't want to invest hours just to get a build system working. >- the state of programs needed for a 1.0 release > - if they build, don't build, haven't tried, what needs to be done good idea >- documentation > - Andreas has already laid out a filesystem framework with unixos2_fhs.txt I hope the GCC maintainers will pick that up soon. > - porting guidelines/logistics needs some up to date documentation. >It would be interesting to see if something like this could be compiled >so it could be made available via unixos2.com. Perhaps a way that folks >could see what others are working on so we don't double up. This may >also help give a more easily quantifiable way of seeing where we are in >the whole scheme of things. I still have some problems with unixos2.com and os2ports.com. IMHO the main target of those pages should be to provide direct information about the current state of unixos2 and also some porting guides. It does not make sense to provide the same information like os2.org or other pages, it should simply consentrate on ports and unixos2 (like LIBEMU, XFree, GCC...) It should also provide direct links about how to set up a unixos2 tree on the system. I know that it's possible to find that if you look long enough but it's not really obvious IMHO. And as long as people have to think before they find the things they need a lot of people won't get it :) I also think that this is the main demotivator right now, several people did something but nothing was really clear. The FTP structure changed several times and people didn't got what to find where now (I have the same problem). IMHO the webpages should contain (at least) the follwing information: (they should be *easy* and *obvious* to find on the page without 1000 indirect links) - what is the idea of unixos2 - FHS, what is it and why does it make sense (i.e. why should all programs take care of UNIXROOT) - compile libraries just once (avoid duplicates) - which packages are available (and where) - where to get support (unixos2 mailinglist) - packages - how to get them and how to install it (especially focus on "bootstrap" packages to get initial core) - list of most recent packages - todo list for potential developers - search function (like search the package with lib... in it) - developers - short development-environment howto - gcc - automake/conf & co - make - tools like awk, sed... - longer howto's (Imakefile & co, done by others already) - simple samples about how to compile on OS/2 - lookout about LIBEMU :-) - ... ? I am willing to help on that but not before january 2002 because I write my diploma work right now. Who is maintaining which page right now? What's your opinion about what those pages should provide? What do you think about my ideas ( I'm not the only one I guess, Jeff mentioned exactly those points)? cu Adrian -- Adrian Gschwend at netlabs.org ktk at netlabs.org ------- Free Software for OS/2 and eCS http://www.netlabs.org **= Email 5 ==========================** Date: Mon, 28 Oct 2002 19:23:18 +0100 From: Thomas Hoffmann Subject: Re: R port? Dies ist eine mehrteilige Nachricht im MIME-Format. --------------D6FD3F289626DC6096AE9BEA Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Pete, I am a bit surprised that you did not find any traces of OS/2 related activities in (e.g.) the R mail archives. Looking in my "R porting folder" I find that I did start porting attempts for R around R version 0.64 which was 3 years ago. I did all this porting as a non-official activity and it has always been my impression that multi platform for (some members of) the R core team definitely means restriction to "Unix+Windows+Mac". So one can 1) pretend to be one of those systems (which means "Unix"), 2) maintain a "private" OS/2 tree (and use diff/grep/... and a lot of time to search through the code with every new release) or 3) try to persuade the R core team to add a system specific extension for OS/2. What I did up to know is 2) (with an PM graphics driver modelled after the Gnuplot PM driver and the X11 graphics driver, a working build system, ...). This way I built 0.64, 0.65, 1.2.3, 1.3.0, 1.3.1 and 1.4.1. On request of Alex Newman (who sometimes seems to read this list) I did package a binary version of R-1.4.1 and mail it to him. All these versions were build with an "traditional" tool set (gcc 2.8.1, autoconf 2.13, a patched posix2 libcext: anybody remembers this library? ;-), ...). But then I decided to switch to a more current environment and started experimenting with gcc 3.0.3, make 3.79.2, autoconf 2.53, ... for building R-1.5.1. (R started requiring autoconf 2.5+ around this time). Well, before I could build a first binary version, R-1.6.0 was out. Now I am still struggling with this new environment (will try the autoconf hint from Thomas Dickey then: the failing autoconf code sample in the recent mail was from R-1.5.1). I am not sure if I will be able to maintain R further (and never was, that's why I did not publish anything, besides, for my 33kbit line R is quite big). I tried to standardize on tools that are current and actively maintained for R porting, but gcc, posix2, unixos2, (even Perl, X11 and the legendary emuos2) are examples of software that seems to dry out on OS/2 (only autoconf and make are clear counter examples). The responsiveness or mere visibility of some maintainers is too small: So gcc 3.0.3 has a strange habit of output filename creation, I ask a question about this here and the answer is zilch. (See the "Roadmap" thread on this list for another example) So there is a real chance that I will never come up with a public release of an OS/2 version of R: simply next to nobody is interested in taking part in porting R to OS/2 (for "R" insert a software name of your liking :-((). Enough whining, my advice: if you are really interested in R-OS/2 than ask me for porting hints and do not try to start from scratch. I will try to help you as much as I can, but a current binary package from me will be some minor versions away. Thomas. Pete Milne schrieb: > > Part 1.1 Type: Plain Text (text/plain) > Encoding: 7bit but the version we have at UEA is 1.5.1 http://cran.r-project.org/src/base/R-1.5.1.tgz (5.2Mb) which expands to about 25Mb. At the moment I'm so short of HD space that even quite straightforward builds fail. I'm currently looking around for an additional SCSI hard disk, meanwhile if I can free up enough space I will attempt to build R-1.5.1. I'm a bit surprised to find no record of R (or its predecessor S-plus) ever being ported to OS/2. -- Thomas Hoffmann Telephone: 49-351-4598831 thoffman at zappa.sax.de Dresden, Germany ..sig under construction ... --------------D6FD3F289626DC6096AE9BEA Content-Type: message/rfc822 Content-Transfer-Encoding: 7bit Content-Disposition: inline >From eyup.org!owner-os2-unix Mon Oct 28 18:25:08 2002 Received: from eyup.org by zappa.sax.de (UUPC/extended 1.13k) with UUCP for thoffman at zappa.sax.de; Mon, 28 Oct 2002 18:25:08 +0100 Received: from mail.eyup.org (213-152-37-93.dsl.eclipse.net.uk [213.152.37.93]) by sax.sax.de (8.9.3/8.9.3) with SMTP id OAA17262 for ; Fri, 25 Oct 2002 14:10:48 +0200 (CEST) Received: by mail.eyup.org (IBM OS/2 SENDMAIL VERSION 2.03/2.0) id MAA431.76; Fri, 25 Oct 2002 12:49:27 +0100 Received: from cmailg6.svr.pol.co.uk by mail.eyup.org (IBM OS/2 SENDMAIL VERSION 2.03/2.0) id MAA431.72; Fri, 25 Oct 2002 12:49:24 +0100 Received: from modem-253.north-dakota.dialup.pol.co.uk ([62.137.85.253] helo=milnefamily.freeserve.co.uk) by cmailg6.svr.pol.co.uk with esmtp (Exim 3.35 #1) id 1852xp-0006dz-00 for os2-unix at eyup.org; Fri, 25 Oct 2002 12:49:19 +0100 Message-ID: <3DB92ECF.57AED539 at milnefamily.freeserve.co.uk> Date: Fri, 25 Oct 2002 12:45:20 +0100 From: Pete Milne X-Mailer: Mozilla 4.61 [en] (OS/2; U) X-Accept-Language: en MIME-Version: 1.0 To: os2-unix at eyup.org Subject: Re: R port? References: <3DB67C02.C928E4C1 at milnefamily.freeserve.co.uk> Content-Type: multipart/alternative; boundary="------------A88167EB6720F96116A6980C" Sender: owner-os2-unix at eyup.org Precedence: bulk Reply-To: os2-unix at eyup.org X-Mozilla-Status2: 00000000 --------------A88167EB6720F96116A6980C Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sorry, the link in my original post is dead. It is The R Project for Statistical Computing and source is available from http://cran.R-project.org The latest version is http://cran.R-project.org/src/base/R-1.6.0.tgz (8.2Mb) but the version we have at UEA is 1.5.1 http://cran.r-project.org/src/base/R-1.5.1.tgz (5.2Mb) which expands to about 25Mb. At the moment I'm so short of HD space that even quite straightforward builds fail. I'm currently looking around for an additional SCSI hard disk, meanwhile if I can free up enough space I will attempt to build R-1.5.1. I'm a bit surprised to find no record of R (or its predecessor S-plus) ever being ported to OS/2. Pete --------------A88167EB6720F96116A6980C Content-Type: text/html; charset=us-ascii Content-Transfer-Encoding: 7bit  
Sorry, the link in my original post is dead. It is
 The R Project for Statistical Computing 

and source is available from  http://cran.R-project.org

The latest version is  http://cran.R-project.org/src/base/R-1.6.0.tgz (8.2Mb)

but the version we have at UEA is 1.5.1
http://cran.r-project.org/src/base/R-1.5.1.tgz  (5.2Mb)
which expands to about 25Mb.
At the moment I'm so short of HD space that even quite straightforward builds fail. I'm currently looking around for an additional SCSI hard disk, meanwhile if I can free up enough space I will attempt to build R-1.5.1.

I'm a bit surprised to find no record of R (or its predecessor S-plus)  ever being ported to OS/2.

Pete --------------A88167EB6720F96116A6980C-- --------------D6FD3F289626DC6096AE9BEA-- **= Email 6 ==========================** Date: Mon, 28 Oct 2002 22:25:15 +1100 (EDT) From: "IanM" Subject: Re: Roadmap? Hi John >After some initial debate many months ago, we had adopted Slackware as >Linux distro to model UnixOS/2 on, and I spent around six months putting >together OS/2 equivalents of around 90 Slackware packages, using the most >uptodate OS/2 copies of any binaries I could find. This was all in place >on unixos2.com at one time, but may well have disappeared. I'm missing some stuff ? If I am, tell me and I'll see about getting it (back). >Over time it seems to have become apparent that no one knows how to use >Slackware pkgs anyway. Maybe people were expecting to be able to download >a complete distro and have it up and running on the click of an icon... I think they expect to have there hands held to much sometimes. :-( But personally I think what you have done is great but.... >People forget that the whole UnixOS/2 distro is largely work in >progress and has been anywhere near a v1.0 release, as a single package. >It also emerged that Slackware was not seem very FHS compliant. The Slackware was a good idea at the time but both myself and others have found it confusing. You did take leadership though and it had to start somewere ! >Since the assembly of the distro didn't seem to be making any progress, I >thought it would be a good idea to try and define a basic toolset which >could be used to build packages from source and this did seem to work very >well for the purpose of building Perl in a 'clean' environment. As a >result of quite a lot of feedback and some excellent contributions, we did >manage to come up with a method of building Perl v5.8.0 for OS/2 on the >day the source was released. IMV that was a major achievement, and it >would be great if we could do that with all Open Source Software without >requiring the intervention of the OS/2 maintainer to do his bit. I'd like >to see the same situation with the likes of gcc, emacs and all the other >big programs that we are fmiliar with. This is were we need "maintainers" to submit OS/2 specific changes to the original code maintainers, and is also part of the stumbling block, not all original code maintainers are interested in going to the trouble to add OS/2, or even Windows specific changes, when they do though, thats great. The main problem I've come across over the past few years in trying to compile packages (and I dont claim to be an expert) is were the OS/2 specific changes included in the porters code that has been included with the earlier binaries appears to be incomplete, missing a diff file, and lack of documentation as to what had to be done. I can understand the documentation part, as its not always easy to remember what was done but generating a diff file against the original code would make life so much easier (as some do thank goodness) and would be a life saver for some projects. >Some months ago, I proposed a UnixOS/2 build system which would >incorporate all OS/2 ports of Unix programs, and this would allow users to >build binaries for themselves using a standard BUILD program. This would >'empower' users to embed whatever drive letters, if any, in their programs >instead of complaining that they were built with 'C:' and their own >UnixOS/2 system was on 'F:'. The basis of that Build System was used for >building Perl, and I was hoping to extend it, but didn't get much >feedback. I think having a standard build system is essential part of >making any progress towards a UnixOS/2 distro because as things stand one >porter may develop a program which works fine for him but has a >prerequisite for a particular 'something' which other people don't have so >the program won't work. The porter may not even be aware of of what that So true but on AIX I've also found this a problem, just not to the same extent. >'something' is. It may be an environment variable, a very specific version >of a particular DLL, something in the PATH, or maybe a thousand other >things, but without a standard build environment it is impossible to >guarantee that a program will build or work, on two different machines. So far, I've managed to successfully compile 2 packages out of about 20 under OS/2, makes me feel like a completely hopeless case, I also spent many a long night trying to figure out what was wrong with my setup, only to give up in frustration. >A baseline toolset is something which I have suggested as a starting point >for the build system, and as was shown by the build of Perl, this can >ensure that people can manage to build something quite complex with little >or no previous experience of building anything from scratch. The results >that several people submitted were extremely similar and it would have >been useful to carry on from that point. >Maybe I should try an get my proposed Build Sytem uploaded to unixos2.com >so that it can be tested more fully... Um, yes please, I was going to query you re your build system :-) Part of the stalling might also be related to the undated GCC, I havent make any attempts to recompile anything myself for a while, as I've been patiently waiting for the latest comilers to be reasonably stable, or has this happened and I've missed it. Cheers IanM http://www.os2site.com/ Iraqi rifle for sale. Never fired. Dropped once.