home *** CD-ROM | disk | FTP | other *** search
- Path: enews.sgi.com!decwrl!waikato!not-for-mail
- From: sbh@cs.waikato.ac.nz (Stephen B Hodge)
- Newsgroups: comp.sys.amiga.misc
- Subject: Re: AmigaOS 4.x features
- Date: 8 Mar 1996 02:51:21 GMT
- Organization: The University of Waikato
- Sender: sbh@waikato.ac.nz
- Message-ID: <4ho7b9$2786@thebes.waikato.ac.nz>
- References: <4h2p87$nu@news.rhrz.uni-bonn.de> <4hiq13$h6c@thebes.waikato.ac.nz> <4hl0km$e72@news.rhrz.uni-bonn.de>
- Reply-To: sbh@waikato.ac.nz
- NNTP-Posting-Host: borg.cs.waikato.ac.nz
- X-Newsreader: TIN [UNIX 1.3 950726BETA PL0]
-
- Bernhard Fastenrath (fasten@zeus.informatik.uni-bonn.de) wrote:
- : : I say go with Java, especially in a network environment.
- : Hmpf, maybe I will get used to Java but I'm not (yet) convinced ...
- The avantage that Java has, as I see it, is that it is targetted at being a
- sercure networking language. I admit I'm fairly ignorant when it comes to
- Pythons capabilities, but I think having Java built into the OS would allow
- a new level of integration with the net - maybe even to the point of Sun's
- vision of applications made up of network sourced Java applets (not likely
- in the near future, and MS will do their best to prevent that). It would be
- nice to use a Java applet from whatever editor, paint program, or whatever.
- Another possibility is perl which seems to have a particularly fanatic bunch
- of fans (don't know much about perl either).
-
- : : : - a hidden flag for assignments the user doesn't want to see (If
- : : : program XY wants XY: assigned to its home I'm not interested).
- : : Why not implement the hidden protection bit, and allow assigns to have
- : : protection bits (they should have anyway). That way they they can be hidden
- : : as you want, but other files and directories can also be hidden.
- : Sounds good. But please give me a "ls -a".
- Could have a command line switch (probably best for shells), and a preference
- option for requesters.
- People should think of the ability to hide files as an organisational tool,
- not as a pathetic attempt at security.
-
- : : : - Workbench should be multi-threaded (and allow grid positioning
- : : : like ******* ** (no that's not "Clean Up")).
- : : Workbench needs to be made more functional. Things like automatically closing
- : : the parent window (like Macs with option double-click), multiple windows
- : : similar to X Windows virtual window managers, make it more user customizable.
- : There are already several public screen managers which are similar to
- : unix virtual window managers but it's not a bad idea to include one with the OS.
- A virtual windows are not so important when there are easy ways of iconising
- or automatically closing windows, I'm not sure I would bother using a virtual
- window manager for amigaos (I use a unix one), but the feature should be added.
-
- : : : - low level game library (not really an OS feature)
- : : No. Games should go through the OS. The power to make this viable should be
- : : available with PPC, so a special mechanism for games should not be required.
- : Maybe, but most games want to lock out the OS anyway.
- But that is unacceptable. Games must be made to behave as any other program.
-
- : : : - MacOS binary compatibility (ShapeShifter)
- : : No. On a PPC machine, if you want to run mac binaries, you buy MacOS, if you
- : : want to run Windoze you buy WindowsNT, if you want unix you buy one of several
- : : flavours of unix available. We want an Amiga so we should stick to an Amiga.
- : On my A3000 they run both at the same time. I like it but you are probably right.
- Now there's an interesting idea - have the ability to run other OSes in amiga
- screens, so all you have to do is buy AmigaOS, MacOS, WindowsNT and Solaris,
- and AmigaOS gives you the ability to run them all simultaneously. I doubt this
- is technically possible, but it would be an incredible selling point, even if
- people only wanted AmigaOS so they could run Mac and PC programs at the same
- time. I definately don't like the idea of have a computer that has the ability
- to actually run MacOS, but instead using an emulator under another OS - it
- seems like a waste to me.
-
- : : The filesystems need minor improvements, to be useful in a networked
- : : environment, adding owner ids, and different security levels (ie. owner, group,
- : : system protection bits).
- : How about the linux ext2fs? Of course, there should still be a single-user
- : filesystem included with the OS.
- One advantage AmigaOS has is the ability to seemlessly integrate a number of
- filesystems. Lets use it to its full advantage.
-
- : : One thing that would also be nice is a dynamic linking system (which is similar
- : : to our runtime shared libraries, but not quite the same - we should have both).
- : I have heard this before but, strange but true (at least strange), I never missed
- : dynamic linking.
- Its not a big thing, although some people like the idea.
-
- : : I think if all of that could be done, will keeping the OS fast (which is really
- : : only a task scheduling thing), we would have an OS better than any on the
- : : market at the moment. It is probably better to forget about compatability, and
- : : provide that as either as seperate OS (just a port of 3.1) or as a set of
- : : libraries - ie. have a new executable code format / task structure so that
- : : the OS can give a program the libraries it wants.
- : Compatiblity is never the "all or nothing" decision I've often heard in this
- : newsgroup. Compatiblity to what do you want to give up?
- Direct access to the custom chips is what I think will have to go, which will
- break a lot of programs (ok, only the games mainly). Its a hard decision to
- make. The choices I see are:
- 1. Enhance and extend the existing libraries, there by maintaining
- compatiblilty with many existing applications.
- 2. Rewrite all the libraries completely, removing the baggage of backwards
- compatibility (eg many off the old graphics routines, reduce the number of
- memory allocation variants, etc). This has some merits in that 68k binary
- compatibility need not be part of the new OS, by providing a seperate
- miniOS to run older programs (ie. contains the 3.0 or 3.2 libraries, and
- an 68k emulator).
- Its likely that (1) is going to be choosen, although in many ways I would be
- happier with (2).
-
- --
-
- Steve Hodge
- sbh@cs.waikato.ac.nz
- Computer Science Dept, University of Waikato,
- Hamilton, New Zealand
-
-