Re: ANN: the unix pestilence

From: Pascal Bourguignon (spam_at_thalassa.informatimago.com)
Date: 05/13/03 

Date: 13 May 2003 07:26:34 +0200

Derk Gwen <derkgwen@HotPOP.com> writes:

> # Please share a couple of your favorite references!
>
> Unfortunately they would be rather dated. You can start with places like
> http://directory.google.com/Top/Computers/Software/Operating_Systems/Research/

Have a look at eros http://www.eros-os.org/
 
> Things like nanokernels and exokernels that would allow multiple
> operating systems on one machine, a process even farming out tasks
> the operating system that best matches its needs of the moment.
>
> Before unix, multics had the notion that all disk files are attached
> as vm segments instead of openned as files. Access the disk files as
> vm instead of files and you automatically have persistent
> objects. Shared vm, something unices still can't seem to get right,
> was as simple as both processes openning the same file and getting
> the same segment.

O, you know that there is mmap. The philosophy of unix is to have
only the minimum service implemented in the kernel, and have the rest
in user-mode processes and in libraries. Guess what? Unix is THE
micro-kernel. Now you want multics files? Just write them in a
library and have applications who need them link with that library:
you have the mmap primitive you need to implement it!

> Some experimental systems had the notion of instead of openning an
> application to open a file, the document had data segments that
> corresponded to the document text, and other segments would be
> executable code segments that integrated into the object. The
> concepts of file and program become more intermingled.

MacOS also had data forks and resource fork, and "data" files could be
distributed as an application. For example, self extracting archives
are implemented this way. Otherwise, data files could shadow any kind
of resource including code resources (not nice, shadowing
application's code!) or other *DEF resources which contained code for
controls, windows, menus, etc. Imagine all the way you could write a
virus! The mere insertion of a floppy would open a resource file and
do such a shadowing...

Otherwise, gluing code to data is not efficient and can be justified
only when the program is small and the data is big, like for self
extracting archives. Try to copy 12MB of MS-Word in each of your 100
byte memo!

Note that MacOSX (inheriting from NeXTSTEP) has file packages, that
are directories where you can put several files and programs, and that
are show as a single icon. So if you will, you can bundle code+data
in MacOSX file packages. Of course, the same can be done on any unix.
You just have to lose that silly custom of dispating files in
directories such as bin, lib, etc, man, ... Put everything in
/opt/stuff (or in:
~/my-documents/doc1.txt/{text,author,copyright-notice,edit*,preview*,print*}
) 

-- 
__Pascal_Bourguignon__                   http://www.informatimago.com/
----------------------------------------------------------------------
Do not adjust your mind, there is a fault in reality.

Relevant Pages

  • Re: Process Resource Manager (PRM) on HP UNIX
    ... Oracle has RAC and a product called Database Vault for this purpose. ... Oracle has it's own Resource Manager in the database. ... There will be two servers and databases can be set ... HP UNIX PRM is not based on UNIX process priority instead based on ...
    (comp.databases.oracle.server)
  • Sr. Developer w JAVA, PERL, CGI
    ... Server-side GUI web and applications development using Perl, CGIs, ... server side web development in a Unix environment. ... individual needs to be an expert on web software development ... The resource will develop and enhance ...
    (comp.lang.javascript)
  • Re: Alternate Data Streams aka Forks
    ... There was a 32K limit to the size of a resource, ... broken into 32K segments, and loaded into memory by segment, as needed, ... When Apple switched the Mac from the 680x0 processor to the Power PC ... would now be in the data fork. ...
    (comp.sys.mac.system)
  • Re: question about tasks, multithreading and multi-cpu machines
    ... Since Ada had tasking before unix had threads I think Unix got the name wrong:) ... Whether it is a human resource or a computing resource does not matter - what's important is that this "resource" is treated as an *ability* to perform work. ... The tasks in Ada do not match any of the above. ... Other languages also got it wrong, by treating threads as objects or even by allowing a programmer to "subclass" or "specialize" a thread class. ...
    (comp.lang.ada)
  • Re: ANN: the unix pestilence
    ... Things like nanokernels and exokernels that would allow multiple operating systems ... Before unix, multics had the notion that all disk files are attached as vm segments ... was as simple as both processes openning the same file and getting ...
    (comp.unix.programmer)