Return to BSD News archive
Path: sserve!newshost.anu.edu.au!harbinger.cc.monash.edu.au!simtel!news.kei.com!bloom-beacon.mit.edu!spool.mu.edu!agate!biosci!news.Stanford.EDU!nntp-hub2.barrnet.net!news1.digital.com!pa.dec.com!nntpd.lkg.dec.com!usenet From: Jon Jenkins <jenkinsj@ozy.dec.com> Newsgroups: comp.unix.bsd.freebsd.misc Subject: Re: The Future of FreeBSD... Date: 23 Jul 1995 12:32:42 GMT Organization: Digital Equipment Corp Lines: 196 Message-ID: <3utfha$3ll@nntpd.lkg.dec.com> References: <3uktse$d9c@hal.nt.tuwien.ac.at> <3umkok$de2@nntpd.lkg.dec.com> <marcus.197.009F3034@ccelab.iastate.edu> <3us0rg$7ph@nntpd.lkg.dec.com> <3usrgl$9uk@felix.junction.net> NNTP-Posting-Host: ozyd13-p3.ozy.dec.com Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Mailer: Mozilla 1.1N (X11; I; BSD/386 uname failed) X-URL: news:3usrgl$9uk@felix.junction.net Hi Michael, michael@okjunc.junction.net (Michael Dillon) wrote: >In article <3us0rg$7ph@nntpd.lkg.dec.com>, >Jon Jenkins <jenkinsj@ozy.dec.com> wrote: > >>Linux was never advertised and look at its poularity. >>In Europe its user base is 10x that of FreeBSD. > >Wait a minute. Linux is advertised in Europe as it is advertised >elsewhere. I believe the first book published about Linux was in Germany. >It received a lot of positive magazine coverage in Europe. Sorry I was talking about when Linux first started. I can still remember downloading the different bits from all over the place. You are of course correct now that Linux has such a large base it is advertised quite heavily. ..snip > >There's no reason why FreeBSD couldn't do the same. In particular, if >someone were to publish a FreeBSD magazine, the popularity of the system >would grow faster than it already is. And make no mistake, FreeBSD's >popularity is growing. A lot of us who use Linux are evaluating FreeBSD >or have plans to do so. Many people use BOTH systems depending on the job >they need done. Sometimes Linux is better, sometimes FreeBSD is. I suppose it really depends on where FreeBSD community wants to head. If it is technical excellence rather than ease of use then real time, multiprocessor extensions, efficient fs etc are probably the way to go. If however the aim is to make FreeBSD as attractive as possible to potential users without compromising it's compatibility with the historical UNIX base I suggest the best way to do this is to concentrate efforts on ease of use factors and a comprehensive GUI app builder closely followed by apps to do all the nasty stuff would be paramount. My personal preference would be ease of use but others may and will differ. NT is a direct competitor for UNIX and you have got to ask why use UNIX instead of NT? My only rational reason is cost i.e. it (FreeBSD) and most of the tools are free!! A less rational reason is my enjoyment in being able to "see under the hood" and being able to customise my environment i.e. shells, window managers source for the apps etc. In my experinece organisations started using UNIX because ther was no alternative. No other OS provided the extensive range of tools or the OS support for the types of applications that they wanted to run. That is no longer the case both OS/2 and NT provide most if not all the functionality of UNIX and in many areas superior OS strategies and the "power" apps that were once the domain of the RISC workstations are now available on these OSs. The next few years will be very interesting indeed!!! ..snip > >There is no doubt in my mind that FreeBSD is a success and will continue >to be so. There is no doubt in my mind that the existence of TWO free >UNIX workalikes helps both of them. It legitimizes the whole idea of a >freely available O/S. I agree. I would however like to see it's popularity spread into the non academic and non "hackers" realm. I believe the single best way to do this is to have a nice slick GUI for everything. > >This thread started out with GUI as the topic. A lot of people believe >that FreeBSD has a GUI because it has X. This is not true. The term "GUI" >no longer refers to what X does. The term has changed its meaning to >include the typical things that people do with a GUI, the typical >programs they run, the typical ease (or expected ease) of getting things >done. The MacOS is the epitome of a GUI. FreeBSD is nowhere near this. but it could be!! > >However, there is no reason why FreeBSD could not become something much >closer to the epitome of a GUI. Caldera (http://www.caldera.com) has the >right idea in building a network desktop based on a UNIX system under the >hood. They happened to choose Linux. In the past NeXT chose Mach as their >UNIX-like underpinning. Work with OS/2 for a while and you'll see that PM >and SOM are layered on top of a multitasking O/S that is powerful in its >own right. As for Mach so did NT and OSF/1 !! As for Next its a shame they chose the 68k base as its death was also Next's death. i really liked Objective C it is so much cleaner than C++. > >What I'd like to see is a componentware approach to the GUI that builds >upon UNIX's strengths as a "toolbox" O/S with lots of filters and tools >that can be glued together in a myriad of ways the designer never thought >of. I don't believe that anybody has done this well, especially not >Microsoft with their OLE concept, but then I believe it's because few >people outside the UNIX world really understand the simplicity and power >of filters and pipes and scripts. I remember when Windows 3.0 started to >gain market acceptance and people talked about how you could buy a word >processor from one company and a spell checker from another and they >would all work together simply and easily. Instead we have bloated >do-everything applications that take 50 megs of hard drive to install, >are awkward to use, to learn, they crash too often, the design by >committee is inconsistent, the error messages are confusing or stupid.... The OO concept has some advantages after all!! > >Can't we do better? > >>Windsock was trivial, perhaps a few minutes. /etc/hosts >>/etc/hosts.conf, bind, NIS, SLIP, PPP, routing, device >>slattach, etc etc .... There really is no comparison > >I've just been evaluating Winsock's and I must say I've been through hell >trying to get them configured and working properly. This includes >commercial products like PC/TCP and PC/NFS I only speak for myself having set up Trumpet Windosock once each on ether, SLIP and PPP. I had all three systems running under a few minutes. Netscape, Eudora and Ewan followed a few minutes later. It may well be that as a general rule it is much more difficult. > >>efficiently via object based GUI builders. Lets say >>I want to change to look of the filemanager window >>in xfm without changing the functionality: >>I have get in the code and hack several >>thousand lines of code combined with an intricate >>knowledge of X intrinsics. If it was developed with a >>good object based GUI builder I wouold start that up >>and "drag n drop" a few components, point the >>event handlers at the backend functionality >>and recompile and exit. > >No recompile. Just like using ResEdit on a Mac, you should be able to >change some stuff and it's done. The sophisticated end user should be >able to do it even if they don't know how to program or understand a >compiler error. I agree. > >>built. With C++ and perhaps LessTif or the like there is >>no technical bar to something similar for UNIX/X. I know >>that SUN and perhaps SGI are already working on something >>similar based on Tcl/Tk if rumour is correct. > >They are working on making TCL/Tk the GUI/scripting system by some small >redesign for portability and by releasing ports for UNIX, MacOS and >Windows. The TCL/Tk source code will be freely available, but their GUI >builder will be a commercial product. It's a start. A componentware >system needs a scripting language, but it is much more than just a GUI >with a scripting language. Check out comp.lang.tcl for more details or else >http://www.sunlabs.com:80/research/tcl/ will take you to dozens of WWW pages >telling you everything you want to know about TCL and Tk Why do we need a scripting lanquage at all? If you haven't yet seen Borlands Delphi take a look. There really is nothing that even compares either in the MS Windows or the UNIX/X world. As for ease of use and user friendliness (is there such a term?) Tcl/Tk is still way to far off the mark. Both the base AND the App builder need to be OO based otherwise you lose half the benefit of the extensibility, replacability and inheritance of components otherwise we end up in the same "deadly embrace" of reinventing the wheel i.e thousands of lines of code (be it Tcl or C++ or whatever) for every app. Thanks for your reply, Jon -- ---------------------------------------------------------------------- Name: Dr Jon Jenkins Location: Digital Equipment Corp, NaC, Burnett Place, Research Park, Bond University, Gold Coast QLD, AUSTRALIA 4229 Phone: 61-75-75-0151 Fax: 61-75-75-0100 Internet: jenkinsj@ozy.dec.com Close Proximity: "HEY YOU !!!" The opinions expressed above are entirely personal and do not reflect the corporate policy of DEC or the opinions of DEC management. -----------------------------------------------------------------------