From mboxrd@z Thu Jan 1 00:00:00 1970 To: 9fans@cse.psu.edu From: Andrew Zubinski Message-ID: <3A424B38.5ED4ABF6@itc.kiev.ua> Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit References: <91nt3g$t2f$1@nnrp1.deja.com>, <20001220195936.B199@localhost.local>, <91ra7i$54s35$1@ID-64718.news.dfncis.de> Subject: Re: [9fans] Re: Future of Plan9 Date: Thu, 21 Dec 2000 18:27:39 +0000 Topicbox-Message-UUID: 3be9b44a-eac9-11e9-9e20-41e7f4b1d025 > But remember one point: Plan 9 is a system use to serve as a > distributed system, but a personal computing system. If you like the > functions provided by those fancy OSes, by all means change back to it. We > enjoy Plan 9 as it is simple, easy to use (is it really easy for beginners What the nice style to help to beginners... There was my question/proposition about the cheap non-x86 hardware for Plan9 terminal (and portable POSIX implementation of cpu/file servers). Now I'm finished the prototype PCB for such beast (133 MHz 64-bits IDT Embedded Orion, 128 MB of DRAM with some fancy features). But I change my mind and decide to implement microkernel based Oberon instead Plan9. Why ? (It is my point of view and, please, don't blame me.) If you are not a computer scientist - purist you'll need an OS with the highly balanced design, firstly - with simple but feature rich API. Why? Cause you'll never find enough time for learning all thouse bang'n'whistles things like MFC/Motif/CDE/GTK/Gnome/Qt/KDE... . But some features are really needy like graphics subsystem for data visualization (does anybody really like to read a 500 MB datafile with simulation results?) and with enough interactivity (o'k, you can write script which will generate postscript from datafile, then render ps and view it, but how to change one or more values in this "small" source datafile?). What I like in Plan9 design are "components as fileservers" conception and Oberon-like Acme. But there are too many Unix garbage in system design at user level - unstructured man pages set, obsolete formats, nothing-new shell, nothing-new utilities set, C compiler only (and nowbody even don't think about the possibility of porting something like Objective-C preprocessor and runtime - sorry, I'm trying, but under the work pressure have no time). As the result instead personal one-box Unix with TeX (or Lout), Octave, Ghostscript, gv, and what-you-want-how-you-want else, you'll have an equal personal 3-box Plan9 network with... oh, yes, TeX, Ghostscript, something-like-gv and what-you-can-port-with-APE. And the real power of Plan9 are used for nothing - for the Unix emulation. Who really need this "feature" when we have somewhere ugly but stable and very tunable well-documented Unix ? This disbalance between clean and clever ideas on system level and wrong target for common OS design (to build distributed Unix) IMHO is the greatest weakness of Plan9. From the system programmer point of view Plan9 is the great OS, from the application programmer point of view it is even more ugly than Unix. Maybe I'm wrong but I can't see any native Plan9 programm oriented to any non-computer application area where possibilities of Plan9 are usefull (like distributed CAD/EDA, GIS). And system is not too young... So, when I see such posts where somebody tolds us that "it is easy to use" or "change back to other OS", I want to ask: "But how are you using it ? What are you doing with it ? What is your application area ?".