** Old Vintage Operating Systems **
Le Guerrier Louis-Luc
Hello,
I'm presently working on a project about the evolution of operating
systems, and I need to study the early versions of operating systems, which I
haven't yet found.
They are:
CP/M 1.3 (from around 1976): the first CP/M from Gary Kildall
QDOS v0.10 from 1980: the first Seattle DOS version.
86-DOS v0.30 from the end of 1980.
CP/M-86 Preliminary Release from august 1981.
I'm aware these are stored on 8" SSSD disks (except the CP/M-86 which
may come on 5.25"), but I can read the sectors from these disks. I'm not
absolutely looking for the original disk; it may be a copy, an e-mailed
disk image, or the files themselves.
I'll offer a $ reward and postage to anyone who can help me.
Please reply to "legu...@ere.umontreal.ca".
Thank you
Louis-Luc Le Guerrier
Université de Montréal
Montreal, Quebec, Canada
P.S.: Big thanks to anyone who has already helped me to find some
other old operating systems.
timo...@cyberramp.net
legu...@ERE.UMontreal.CA (Le Guerrier Louis-Luc) wrote:
>Hello,
> I'm presently working on a project about the evolution of operating
>systems, and I need to study the early versions of operating systems, which I
>haven't yet found.
>They are:
> CP/M 1.3 (from around 1976): the first CP/M from Gary Kildall
> QDOS v0.10 from 1980: the first Seattle DOS version.
> 86-DOS v0.30 from the end of 1980.
> CP/M-86 Preliminary Release from august 1981.
>
> I'm aware these are stored on 8" SSSD disks (except the CP/M-86 which
>may come on 5.25"), but I can read the sectors from these disks. I'm not
>absolutely looking for the original disk; it may be a copy, an e-mailed
>disk image, or the files themselves.
> I'll offer a $ reward and postage to anyone who can help me.
You might go to my web site and check out the IMDOS binary file there.
This is a copy of CP/M 1.3 private labeled IMDOS. I have binary only.I
also have CP/M 1.4, and a CP/M 1 manual.
No reward necessary.
Tim Olmstead
webmaster of the CP/M Unofficial web page
email : timo...@cyberramp.net
MAIN SITE AT : http://cdl.uta.edu/cpm
MIRROR AT : http://www.mathcs.emory.edu/~cfs/cpm
Lisa or Jeff
> They are:
> CP/M 1.3 (from around 1976): the first CP/M from Gary Kildall
> QDOS v0.10 from 1980: the first Seattle DOS version.
> 86-DOS v0.30 from the end of 1980.
> CP/M-86 Preliminary Release from august 1981.
How about 360-DOS, 360-BOS, and 360-OS? And they're far from the earliest.
Robert Schuldenfrei
I was waiting for someone to mention that CP/M is hardly a Vintage OS. Does
anyone know of an OS older than IBSYS?
Lisa or Jeff wrote in message <6gea3u$e...@netaxs.com>...
Alan Bowler
In article <6gg41v$j...@news-central.tiac.net> "Robert Schuldenfrei" <tang...@tiac.net> writes:
>I was waiting for someone to mention that CP/M is hardly a Vintage OS. Does
>anyone know of an OS older than IBSYS?
>
>>
>
>
although it was close. At the time it was refered to as a "monitor",
or perhaps "executive". There is some validity to the claim that
"OS/360 was the first 'operating system'", but I think that is
stretching it, however other earlier systems (e.g. CTSS) probbaly also
should be classed as true OS's. An OS should provide
1) control and access to all I/O
2) dispatching and interrupt handling
3) task creation and program loading
4) a file system (OS/360 just squeaks through on this one)
5) an extensive set of utilities to do routine tasks.
(file copy, create, delete, compilers, linkers, library
managers etc.)
It helps in creating an OS if the hardware provides protection
mechanisms so that the OS can protect itself and other tasks from
runaway programs, but with coding discipline this is not absolutely
needed. RSX-11, early Xenix, and early versions of Wicat's MCS
all provided OS services without hardware protection.
Real OS's came into existence when several hardware advances were
made
1) Disks became large enough that they could be used for storing
data files, and not just the work area for 1 program or
a fast library of important prelinked programs.
2) Memory became large enough that there was room to hold the
executive part of the OS, plus more than 1 user program.
3) CPUs became fast enough that it was worthwhile to dispatch
to a different program while one program was waiting for
and I/O request to complete. This also required I/O hardware
that did not lock up a CPU.
4) Interrupt mechanisms for reporting conditions like I/O completion.
5) CPUs that trapped with an interrupt like mechanism rather than
halting on certain programming errors
Douglas W. Jones,201H MLH,3193350740,3193382879
From article <6ggd7a$85d$1...@nntp1.uunet.ca>,
by atbo...@thinkage.on.ca (Alan Bowler):
> By the definitions I use, it is not clear that IBSYS was an OS,
> although it was close. At the time it was refered to as a "monitor",
> or perhaps "executive". There is some validity to the claim that
> "OS/360 was the first 'operating system'", but I think that is
> stretching it, however other earlier systems (e.g. CTSS) probbaly also
> should be classed as true OS's. An OS should provide
> 1) control and access to all I/O
> 2) dispatching and interrupt handling
> 3) task creation and program loading
> 4) a file system (OS/360 just squeaks through on this one)
> 5) an extensive set of utilities to do routine tasks.
> (file copy, create, delete, compilers, linkers, library
> managers etc.)
The Feranti Atlas computer, circa 1960, had an OS by this definition.
The B 5000, at about the same time, had an OS by this definition.
Systems like CTSS and OS/360 were much later, and indeed, IBSYS, OS/8
and MS/DOS fail this definition and can be swept under the rug as
non operating-systems (the location of the hyphen certainly matters,
although some people certainly don't think these systems operated very
well).
> 1) Disks became large enough that they could be used for storing
> data files, and not just the work area for 1 program or
> a fast library of important prelinked programs.
ATLAS used paging drums, not disks, and used mag tape for file storage,
but it really managed to use mag tape as a third level backing store.
> 2) Memory became large enough that there was room to hold the
> executive part of the OS, plus more than 1 user program.
Both the B 5000 and Atlas machines had virtual memory, so the large main
memory was needed for performance but not for basic feasibility.
Later, around 1967, the Berkeley Timesharing System on the SDS 940
managed to do decent interactive timesharing (with FORTRAN, LISP, CAL
and assembly language support) using only 16 page frames of main memory,
plus a decent paging drum. The word size was 24 bits.
Not long after 1967, TSS/8 came into the world, running with a maximum
memory of 32K 12 bit words, with swapping to fixed-head disk, and up to
8 interactive users. FOCAL, assembly language, and FORTRAN were
available. Initially, users were limited to 4K of memory each, but
even this was enough for a FORTRAN II compiler!
> 3) CPUs became fast enough that it was worthwhile to dispatch
> to a different program while one program was waiting for
> and I/O request to complete. This also required I/O hardware
> that did not lock up a CPU.
This was all true with many early transistorized computers from the
early 1960s. Certainly, it was true of the B 5000 and the Atlas.
> 4) Interrupt mechanisms for reporting conditions like I/O completion.
> 5) CPUs that trapped with an interrupt like mechanism rather than
> halting on certain programming errors
These were commonplace in new designs of the 1960's.
Doug Jones
jo...@cs.uiowa.edu
Tom Van Vleck
Douglas W. Jones wrote:
> The Feranti Atlas computer, circa 1960, had an OS by this definition.
> The B 5000, at about the same time, had an OS by this definition.
>
> Systems like CTSS and OS/360 were much later,
CTSS was demo'd in 1961.. not *much* later.
Myself I would say that an "operating system" should really include
- Protection of the monitor from user programs
- Protection of user programs from each other
- Limits on resource consumption by user programs
In which case CTSS's date would be 1963, when it was moved to the
two-core-with-RPQs 7094.
What about the BBN timesharing system, circa 1961?
Robert Schuldenfrei
Thanks for the insight Alan. I never used IBSYS, but I do remember it being
called a Monitor. As you point out, what is, or is not an Operating System
is food for debate. Where better to have that kind of cat fight than in
good ole a.f.c. What was the first OS to be CALLED an Operating System?
Alan Bowler wrote in message <6ggd7a$85d$1...@nntp1.uunet.ca>...
>In article <6gg41v$j...@news-central.tiac.net> "Robert Schuldenfrei"
<tang...@tiac.net> writes:
>>I was waiting for someone to mention that CP/M is hardly a Vintage OS.
Does
>>anyone know of an OS older than IBSYS?
>>
>>>
>>>How about 360-DOS, 360-BOS, and 360-OS? And they're far from the
earliest.
>>
>>
>By the definitions I use, it is not clear that IBSYS was an OS,
>although it was close. At the time it was refered to as a "monitor",
>or perhaps "executive". There is some validity to the claim that
>"OS/360 was the first 'operating system'", but I think that is
>stretching it, however other earlier systems (e.g. CTSS) probbaly also
>should be classed as true OS's. An OS should provide
> 1) control and access to all I/O
Clearly something was doing this on the 70x machines. FORTRAN did a fair
job of calling ALL I/O. Since it had control of the whole machine, access
was not a big issue. If IBSYS was "in town", rather than just the FORTRAN
monitor, who did I/O?
> 2) dispatching and interrupt handling
My guess was that IBSYS did dispatching, such as it was. Your application
(FORTRAN, ASSEMBLER, etc) probably had to terminate. In extremis, the
operator was the interupt of last resort. Where is the historical record
did interupts originate? Beyond first use, when did they become part of the
normal way programs and I/O were controlled? Let me guess that Whirlwind
used interupts, but that normal usage was a much later development.
> 3) task creation and program loading
Program loading was probably one of the primary uses of IBSYS, no? That
together with the spooling of I/O to tape and what later became known as
ABEND analysis, was the guts of what it could do.
> 4) a file system (OS/360 just squeaks through on this one)
Yes but file "systems" were not much more than tapes, plus the RAMAC if you
were lucky.
> 5) an extensive set of utilities to do routine tasks.
> (file copy, create, delete, compilers, linkers, library
> managers etc.)
OK, one person's bare minimum, is another person's embarassment of riches,
but there were loaders, assmeblers, and compilers. Add to that dumps and
monitors and you have your rich set of utilities. What did IBSYS have.
Well I probably made enough errors here to keep this group going for a bit.
There are a lot of folks who REALLY know out there. Joe Morris, David Ness,
et. al. - wade in.
Cheers,
Bob
Alan Bowler
In article <6ggj46$hge$1...@flood.weeg.uiowa.edu> jo...@pyrite.cs.uiowa.edu (Douglas W. Jones,201H MLH,3193350740,3193382879) writes:
>From article <6ggd7a$85d$1...@nntp1.uunet.ca>,
>by atbo...@thinkage.on.ca (Alan Bowler):
>
>> although it was close. At the time it was refered to as a "monitor",
>> or perhaps "executive". There is some validity to the claim that
>> "OS/360 was the first 'operating system'", but I think that is
>> stretching it, however other earlier systems (e.g. CTSS) probbaly also
>> should be classed as true OS's. An OS should provide
>> 1) control and access to all I/O
>> (file copy, create, delete, compilers, linkers, library
>> managers etc.)
>
>The B 5000, at about the same time, had an OS by this definition.
I should have been clearer. There were real OSes BEFORE OS/360.
The "some validity" I was attributing was that getting itself and the
term "operating system" widespread (at least on the is side of the
Atlantic). The Atlas and the B 5000 just weren't that well known.
The term certainly got well established and marketing types
proceeded to apply it to many systems (both good and bad ones)
that did not warrant it.
>
>Systems like CTSS and OS/360 were much later, and indeed, IBSYS, OS/8
What was OS/8?
>and MS/DOS fail this definition and can be swept under the rug as
>non operating-systems (the location of the hyphen certainly matters,
>although some people certainly don't think these systems operated very
>well).
>
>> 1) Disks became large enough that they could be used for storing
>> data files, and not just the work area for 1 program or
>> a fast library of important prelinked programs.
>
>ATLAS used paging drums, not disks, and used mag tape for file storage,
>but it really managed to use mag tape as a third level backing store.
>
>> 2) Memory became large enough that there was room to hold the
>> executive part of the OS, plus more than 1 user program.
>
>Both the B 5000 and Atlas machines had virtual memory, so the large main
>memory was needed for performance but not for basic feasibility.
>
>Later, around 1967, the Berkeley Timesharing System on the SDS 940
>managed to do decent interactive timesharing (with FORTRAN, LISP, CAL
>and assembly language support) using only 16 page frames of main memory,
>plus a decent paging drum. The word size was 24 bits.
>
>Not long after 1967, TSS/8 came into the world, running with a maximum
>memory of 32K 12 bit words, with swapping to fixed-head disk, and up to
>8 interactive users. FOCAL, assembly language, and FORTRAN were
>available. Initially, users were limited to 4K of memory each, but
>even this was enough for a FORTRAN II compiler!
1967 makes them contemporaries of OS/360, and the amount of memory you
cite is still in the ballpark of what other later operating systems
needed. It is sad that technically superior systems like the
940 died out, but OS/360 lives on.
>
>> 3) CPUs became fast enough that it was worthwhile to dispatch
>> to a different program while one program was waiting for
>> and I/O request to complete. This also required I/O hardware
>> that did not lock up a CPU.
>
>This was all true with many early transistorized computers from the
>early 1960s. Certainly, it was true of the B 5000 and the Atlas.
I was not claiming it was not. However, there were other hardware
platforms that performed lots of useful computing that did not meet
this requirement. The IBM 1620 comes to mind, the CPU datapaths were
used during I/O transfers so there was no spare CPU to allocate to
another task.
>
>> 4) Interrupt mechanisms for reporting conditions like I/O completion.
>> 5) CPUs that trapped with an interrupt like mechanism rather than
>> halting on certain programming errors
>
>These were commonplace in new designs of the 1960's.
>
I was not claiming they weren't. They were not universal. All I was
trying to say was that you need a certain level of hardware before
having an operating system is feasible or worthwhile.
IBSYS certainly ran on hardware adequate to support an operating
system.
Richard M. Alderson III
In article <6ggo7k$3...@news-central.tiac.net> "Robert Schuldenfrei"
<tang...@tiac.net> writes:
>Thanks for the insight Alan. I never used IBSYS, but I do remember it being
>called a Monitor. As you point out, what is, or is not an Operating System is
>food for debate. Where better to have that kind of cat fight than in good ole
>a.f.c. What was the first OS to be CALLED an Operating System?
I can't answer that question, but I will point out that the Tops-10 and Tops-20
operating systems have always been called "monitors"; I'm not certain of the
terminology for ITS in this regard.
--
Rich Alderson Last LOTS Tops-20 Systems Programmer, 1984-1991
last name @ XKL dot COM Current maintainer, MIT TECO EMACS (v. 170)
Alan Bowler
>
>(FORTRAN, ASSEMBLER, etc) probably had to terminate. In extremis, the
>operator was the interupt of last resort. Where is the historical record
>did interupts originate? Beyond first use, when did they become part of the
>normal way programs and I/O were controlled? Let me guess that Whirlwind
>used interupts, but that normal usage was a much later development.
Did IBSYS support multiple tasks? I never used it, but I always had
the impression that it just loaded and ran jobs sequentially from
the job tape prepared by the local 1401.
>
>> 3) task creation and program loading
>
>
>Program loading was probably one of the primary uses of IBSYS, no? That
>together with the spooling of I/O to tape and what later became known as
>ABEND analysis, was the guts of what it could do.
But I don't beleive it would create a second task, and allocate the CPU
between them.
>
>> 4) a file system (OS/360 just squeaks through on this one)
>
>
>Yes but file "systems" were not much more than tapes, plus the RAMAC if you
>were lucky.
No I mean more than tape I/O. I expect an operating system to manage
actual named files on disk (or drum). It should perform the space
management an hide (or at least smooth out) the details of physcal
location on the disk. (Better systems hide which disk (or disks),
but there are lots of moderen systems that still have troubles
with that.)
Marco S Hyman
atbo...@thinkage.on.ca (Alan Bowler) writes:
> I should have been clearer. There were real OSes BEFORE OS/360.
> The "some validity" I was attributing was that getting itself and the
> term "operating system" widespread (at least on the is side of the
> Atlantic). The Atlas and the B 5000 just weren't that well known.
The Bxxx machines had a control program (the Master Control Program, MCP)
not an operating system. I don't know if that term originated with the
B5000 or later.
// marc
Sarr J. Blumson
In article <6ggj46$hge$1...@flood.weeg.uiowa.edu>,
Douglas W. Jones,201H MLH,3193350740,3193382879 <jo...@pyrite.cs.uiowa.edu> wrote:
>From article <6ggd7a$85d$1...@nntp1.uunet.ca>,
>by atbo...@thinkage.on.ca (Alan Bowler):
>
>managed to do decent interactive timesharing (with FORTRAN, LISP, CAL
>and assembly language support) using only 16 page frames of main memory,
>plus a decent paging drum. The word size was 24 bits.
Earlier, in 1964 the Dartmouth Time Sharing System managed to do decent
interactive timesharing doing full swapping on a big 6 mb (well it was big,
it weighed well over a ton) moving head disk. It used a master/slave CPU
set up to protect the core OS from user programs (which were all written in
allegedly safe higher level languages anyway).
--
--------
Sarr Blumson sa...@umich.edu
voice: +1 734 764 0253 home: +1 734 665 9591
ITD, University of Michigan http://www-personal.umich.edu/~sarr/
519 W William, Ann Arbor, MI 48103-4943
lis...@zetnet.co.uk
On 1998-04-08 tang...@tiac.net said:
:Newsgroups: alt.folklore.computers
:I was waiting for someone to mention that CP/M is hardly a Vintage
:OS. Does anyone know of an OS older than IBSYS?
george, for the elliott computers?
--
Communa (together) we remember... we'll see you falling
you know soft spoken changes nothing to sing within her...
Tom Van Vleck
> On 1998-04-08 tang...@tiac.net said:
> :Newsgroups: alt.folklore.computers
> :I was waiting for someone to mention that CP/M is hardly a Vintage
> :OS. Does anyone know of an OS older than IBSYS?
I think FMS preceded IBSYS for the 7090 series.
For some definition of "operating system."
The University of Michigan ran UMES on its 7090, University of Michigan
Executive System, derived from a GM Research Center executive
that was one of the first operating systems, according to Mike Alexander.
This should take us back to the late 50s.
dave porter
GEORGE was the O/S for the ICL (formerly ICT) 1900
series of machines. I don't *think* it has anything
to do with Elliot.
Mind you, the only Elliot machine I ever used was
a 903 in a basement, and it didn't have any operating
system, nor any rotating magnetic storage either. You loaded your
compiler from paper tape, ran the Algol source tape through
twice (2-pass compiler, obviously :-), and then started
at address 8191.
Then you carefully powered down the machine so that
the core memory would still contain your prgram when
you came in the next morning, so that you didn't have
to mess around with the bloody temperamental paper tape
reader.
Tell that to t'kids of today and they won't believe
you. "Paper tape punch? Luxury! We 'ad to chew 'oles
in the tape wi' our teeth!". Etc.
dave
--
For email, please remove the 'w' from my address. Sorry.
lis...@zetnet.co.uk wrote in article <6gj7s0$l4i$1...@irk.zetnet.co.uk>...
>
>
> On 1998-04-08 tang...@tiac.net said:
> :Newsgroups: alt.folklore.computers
> :I was waiting for someone to mention that CP/M is hardly a Vintage
> :OS. Does anyone know of an OS older than IBSYS?
>
Joel C. Ewing
Alan Bowler wrote:
>
> In article <6gg41v$j...@news-central.tiac.net> "Robert Schuldenfrei" <tang...@tiac.net> writes:
> >anyone know of an OS older than IBSYS?
> >>How about 360-DOS, 360-BOS, and 360-OS? And they're far from the earliest.
> although it was close. At the time it was refered to as a "monitor",
> or perhaps "executive". There is some validity to the claim that
> "OS/360 was the first 'operating system'", but I think that is
> stretching it, however other earlier systems (e.g. CTSS) probbaly also
> should be classed as true OS's. An OS should provide
> 1) control and access to all I/O
> (file copy, create, delete, compilers, linkers, library
> managers etc.)
IMHO, to nail down specific requirements that must be present to be
"worthy" of the title "Operating System" is to miss part of the essence
of the concept: namely that an Operating System converts a real hardware
architecture into a virtual architecture which is a more useful platform
for application development and execution. The specific features
required to do this in a useful manner depend on both hardware and
application complexity. As you and other point out, there has been
considerable evolution in the tasks and features expected in an
Operating System as hardware and software has evolved, but to try to
ascertain the precise points in the evolutionary process to distinguish
the concepts of supervisor, executive, monitor, and "operating system"
is probably not nearly as fruitful as viewing the evolutionary pattern
of support in such areas as your (1)-(5) above. One can see the
beginnings of all of these features even in something as primative as
the IBM 1620 Monitor I & II supervisors, and for the hardware/software
complexity of the time they were quite adequate. All the popular IBM PC
operating systems could be argued as as being too deficient in your
points (1)-(5) to warrant being called operating systems, yet this would
be at variance with common accepted usage.
As an old mainframer (but user of PC platforms), I would of course
concur with anyone who might point out the many inadequacies of the
popular PC "operating systems" when compared with the features and rigor
customarily expected today on a mainframe Operating System; I just
wouldn't concur that they should lose the Operating System title.
--
Joel C. Ewing, Fort Smith, AR jce...@acm.org
Andrew Gabriel
In article <01bd64d3$578d0a60$0ba17392@glastonbury>,
"dave porter" <por...@wultranet.com> writes:
>GEORGE was the O/S for the ICL (formerly ICT) 1900
>series of machines. I don't *think* it has anything
>to do with Elliot.
Relative geneology of ICT/ICL and Elloitt are shown at
http://www.cucumber.co.uk/geccl/19471972/index.html
from which I think you are right. However, some Elliott
and English Electric staff moved over to ICT in the
various mergers/splits.
--
Andrew Gabriel
Consultant Software Engineer
David Green
On 10 Apr 1998 22:53:09 GMT, "dave porter" <por...@wultranet.com>
wrote:
>GEORGE was the O/S for the ICL (formerly ICT) 1900
>series of machines. I don't *think* it has anything
>to do with Elliot.
>
My understanding is that GEORGE dates back to 1957; stood for GEneral
ORder GEnerator; and was developed on an English Electric Deuce
computer at the University of New South Wales in Sydney, Australia
by Dr Charles Hamblin.
It was later adopted by English Electric for its second generation
computer called KDF9. English Electric later merged with ICT giving
rise to ICL. I think it was probably a programming language rather
than an operating system, and that there were both interpreters and
compilers for it.
John Hughes
lis...@zetnet.co.uk writes:
> On 1998-04-08 tang...@tiac.net said:
> :Newsgroups: alt.folklore.computers
> :I was waiting for someone to mention that CP/M is hardly a Vintage
> :OS. Does anyone know of an OS older than IBSYS?
>
> george, for the elliott computers?
George was ICT wasn't it? Considerably older than IBSYS?
--
John Hughes <jo...@AtlanTech.COM>,
Atlantic Technologies Inc. Tel: +33-1-43204546
24 rue Montbrun, Fax: +33-1-43204579
75014 PARIS.
Robert Billing
In article <01bd64d3$578d0a60$0ba17392@glastonbury>
por...@wultranet.com "dave porter" writes:
> GEORGE was the O/S for the ICL (formerly ICT) 1900
True, I still have the manuals.
> system, nor any rotating magnetic storage either. You loaded your
> compiler from paper tape, ran the Algol source tape through
> twice (2-pass compiler, obviously :-), and then started
> at address 8191.
Ahh, smell the nostalgia. Remember the big grey panel with the nice
solid square button that put the reader on line?
--
I am Robert Billing, Christian, inventor, traveller, cook and animal
lover, I live near 0:46W 51:22N. http://www.tnglwood.demon.co.uk/
"Bother," said Pooh, "Eeyore, ready two photon torpedoes and lock
phasers on the Heffalump, Piglet, meet me in transporter room three"
Will Rose
dave porter (por...@wultranet.com) wrote:
: GEORGE was the O/S for the ICL (formerly ICT) 1900
: series of machines. I don't *think* it has anything
: to do with Elliot.
: Mind you, the only Elliot machine I ever used was
: a 903 in a basement, and it didn't have any operating
: system, nor any rotating magnetic storage either. You loaded your
: compiler from paper tape, ran the Algol source tape through
: twice (2-pass compiler, obviously :-), and then started
: at address 8191.
: Then you carefully powered down the machine so that
: the core memory would still contain your prgram when
: you came in the next morning, so that you didn't have
: to mess around with the bloody temperamental paper tape
: reader.
The 903 was the first portable I ever used; in fact, the first
computer I every programmed. It was portable in the sense it needed
a moving van to carry it and its generator around, but it did move.
(I think I've still got the tapes somewhere; a year or so ago I ran
the programs on a emulator, and they still worked).
jmfb...@ma.ultranet.com
In article <ALDERSON.9...@netcom.netcom.com>,
alde...@netcom.netcom.com (Richard M. Alderson III) wrote:
>In article <6ggo7k$3...@news-central.tiac.net> "Robert Schuldenfrei"
><tang...@tiac.net> writes:
>
>>remember it being called a Monitor. As you point out, what
>>is, or is not an Operating System is food for debate. Where
>>better to have that kind of cat fight than in good ole
>>a.f.c. What was the first OS to be CALLED an Operating System?
>
>I can't answer that question, but I will point out that the
>Tops-10 and Tops-20 operating systems have always been
>called "monitors"; I'm not certain of the terminology
>for ITS in this regard.
Well, that's not quite true. The TOPS-10 and TOPS-20
operating systems included the monitor and the user
mode programs. Without those user mode programs,
timesharing wouldn't be quite as interesting or
useful :-). Sometimes, our use of the term operating
system would also include the hardware it ran on. It
really depended on the context of the conversation.
/BAH
Robert Schuldenfrei
It is interesting to note as an American, how much really good and novel
ideas came from the UK. The works of Turing and Wilkes not withstanding,
who are the folks who came up with George? I consider myself somewhat a
student of computer history, but I guess I should make that American
computer history :)
As I stated earlier, I really do not know what IBSYS could do, let alone
George. Could someone post a feature listing of both systems? I once had
an introduction to OS/360 that was printed BEFORE it was released. It has
long since been canned. It would be instructive to also list the features
of it, the actual features of DOS (the mainframe "stop gap" system that was
released as a substitute for OS/360), and then OS/360 as released. It would
be really great if IBM published on the web, those early manuals and
brochures.
John Hughes wrote in message ...
>lis...@zetnet.co.uk writes:
>
>> On 1998-04-08 tang...@tiac.net said:
>> :Newsgroups: alt.folklore.computers
>> :I was waiting for someone to mention that CP/M is hardly a Vintage
>> :OS. Does anyone know of an OS older than IBSYS?
>>
>> george, for the elliott computers?
>
>George was ICT wasn't it? Considerably older than IBSYS?
>
Does anyone have the dates for George and IBSYS?
Robert Billing
In article <352f738b...@news.m.iinet.net.au>
dgr...@iinet.net.au "David Green" writes:
> rise to ICL. I think it was probably a programming language rather
> than an operating system, and that there were both interpreters and
> compilers for it.
I have a George manual, ICL Part number 4345(1.76). It is a loose
collection of pages about one inch thick. I can't find a date on it but
the 76 may be the year. The paper size is A4, which fixes it to be
after the mid 60s, which is when the UK went over from the old imperial
paper sizes to "A" sizes.
Certainly a 1904 and a 1907 were in use at the installation where I
suspect this manual originated in 1976.
The preface begins:
"GEORGE 3 and 4 are the operating systems for the larger computers of
the 1900 series."
It then goes on to describe the minimum configuration:
48K words of core[1]
2M words of direct access backing store
[1] This would have been 48K x 36 bits or just over 200K bytes
Zulkiflii Hamzah
Hello there,
Can somebody recommend / advice me about
a good starting line to write a compiler?
Could be book title, or... maybe some pointer?
fyi I have developed several applications in assembly language.
and, fairly OK-to-competitive in C programming.
Thank you very much.
Zul Hamzah.
timo...@cyberramp.net
Zulkiflii Hamzah <zu...@pc.jaring.my> wrote:
>Hello there,
>Can somebody recommend / advice me about
>a good starting line to write a compiler?
>Could be book title, or... maybe some pointer?
>fyi I have developed several applications in assembly language.
>and, fairly OK-to-competitive in C programming.
HI. If you're wanting to write a compiler f4rom scratch, and how
little or no experience, I would suggest that you consider going the
YACC/LEX route. This will still not reduce the problem to trivial
proportions, but will help. You will have to create a "syntax" for
yacc, then compile it.
YACC and LEX are compiler writing toos that come to us from the unix
world. they are both still very popular in the unix comunity, but
therre are versions available for the MS-DOS world.
A pretty good book that describes these tools is "lex*yacc" published
by O'Reilly & Associates; ISBN 1-56592-000-7.
Probably the biggest disadvantage to this approach is that it results
in a compiler that is usualy far from optimal.
I would also suggest that you check out the DJGPP newsgroup. I don't
have the complete usenet address handy, but you can search for DJGPP
and find it pretty fast. There are a number of people on that ng who
write compilers, and are willing to help. You canm also get the source
to the GNU/DJGPP compilers and play with them.
Good luck.
Sarr J. Blumson
In article <353212...@pc.jaring.my>,
Zulkiflii Hamzah <zulh_n...@pc.jaring.my> wrote:
>Hello there,
>
>Can somebody recommend / advice me about
>a good starting line to write a compiler?
>Could be book title, or... maybe some pointer?
The classic book is by Aho and Ullmann, "Principles of Compiler Design". You
might also want to look at the newsgroup comp.compilers.
Kevin McQuiggin
Another classic reference is "Compiler Construction for Digital
Computers" by David Gries.
Zulkiflii Hamzah wrote:
>
> Hello there,
>
> Can somebody recommend / advice me about
> a good starting line to write a compiler?
> Could be book title, or... maybe some pointer?
>
> fyi I have developed several applications in assembly language.
> and, fairly OK-to-competitive in C programming.
>
Will Hartung
sa...@engin.umich.edu (Sarr J. Blumson) writes:
>In article <353212...@pc.jaring.my>,
>Zulkiflii Hamzah <zulh_n...@pc.jaring.my> wrote:
>>Hello there,
>>
>>Can somebody recommend / advice me about
>>a good starting line to write a compiler?
>>Could be book title, or... maybe some pointer?
>The classic book is by Aho and Ullmann, "Principles of Compiler Design". You
>might also want to look at the newsgroup comp.compilers.
It's a good book. But it may be a bit more "theoretical" than you're
looking for.
What's somewhat amusing is that no one has asked what kind of language
you want to compile. I think it is fair that you're not interested in
PL/I or COBOL or SIMULA. You may be thinking of something of your own
creation, or perhaps a current language.
Some languages may be easier to implement and compile than others, for
various reasons. Some may move the amount of work necessary from the
parser to the code generator and vise-a-versa.
For example, parsing 'C' is probably more daunting that writing the
code generator for it (at least a trivial code generator). FORTH is
absolutely brain dead to lex, nearly so to parse, and actually pretty
trivial to generate code for. LISP can be easy to lex and parse, but
difficult to write an _efficient_ code generator, plus the assorted
system support it needs (specifically garbage collection).
Anyway, there are several "practical" books on compilers, any decent
technical bookstore should have a couple. Many focus on 'C'. There are
several old magazine articles that describe the process, and implement
little languages. And, I imagine that there are several compilers and
interpreters on the 'net for you to look at. Go for the simple ones
(i.e. going through the GCC source isn't recommended for your first
exercise).
Take a look at Amazon.com to get a list of compiler/interpreter books
that you may be interested in shopping for.
Good Luck!
--
Will Hartung - Rancho Santa Margarita. It's a dry heat. vfr...@netcom.com
1990 VFR750 - VFR=Very Red "Ho, HaHa, Dodge, Parry, Spin, HA! THRUST!"
1993 Explorer - Cage? Hell, it's a prison. -D. Duck
Heinz W. Wiggeshoff
Kevin McQuiggin (mcqu...@sfu.ca) writes:
> Another classic reference is "Compiler Construction for Digital
> Computers" by David Gries.
Even when new, my copy was hard on the eyes. It was printed from
a photoreduced printout made with a 1403 running a TN chain.
But if you ever wanted to write an OS/360 object deck, (the primary
input to the linkage editor or loader), most of the painful details
are there.
Alas, nothing about LR(n), LALR(n) etc.
Alistair J. R. Young
On Mon, 13 Apr 1998 21:28:21 +0800, in message <353212...@pc.jaring.my>,
Zulkiflii Hamzah <zu...@pc.jaring.my> (== zulh)
praised Shub-Internet thus:
> Hello there,
> Can somebody recommend / advice me about
> a good starting line to write a compiler?
> Could be book title, or... maybe some pointer?
Along similar lines, could anyone recommend a good starting point for
someone interested in creating an operating system from more-or-less
scratch (other than reading through, say, the Linux kernel source,
etc., which I've already thought of)?
Perhaps something along the lines of "Operating System Design for
Dummies^WReasonably Clueful People Who However Have Only Previously
Written Application-Type Stuff"?
Thanks in advance,
Alistair
--
Computational Thaumaturge -- Sysimperator, dominus retis deusque machinarum.
e-mail: avata...@arkane.demon.co.uk WWW: http://www.arkane.demon.co.uk/
"I will make the main entrance to my fortress standard-sized. While elaborate
60-foot high double-doors definitely impress the masses, they are hard to
close quickly in an emergency." -- #113, The Evil Overlord List
Will Hartung
Alistair J. R. Young <avatar...@arkane.demon.co.uk> writes:
>Along similar lines, could anyone recommend a good starting point for
>someone interested in creating an operating system from more-or-less
>scratch (other than reading through, say, the Linux kernel source,
>etc., which I've already thought of)?
>Perhaps something along the lines of "Operating System Design for
>Dummies^WReasonably Clueful People Who However Have Only Previously
>Written Application-Type Stuff"?
There are several issues relating to OS design, some of which generate
1000's of pages of research papers. There are also several different
aspects to OS design, and not all of them are pertinent to any
specific project.
For generic OS related issues and discussions on the major aspects of
OS, there are a couple of OS texts designed, more or less, for
coollege courses that have chapters on each major subsytem. No real
code, but comparisons of different OS's. I can't think of the name,
however.
Another approach is to look at the either the 4.4 BSD internals book,
or the "Secret Garden" SYS Vr4 book. These go into detail about how
their UNIX kernels were designed and provide semi-source code showing
how the major algorithms looks. I understand that there is a Linux
Internals book floating around as well, but I haven't seen it.
There is also a book titled something like "Create your own 32-bit OS"
where a fellow details his experience crafting one from scratch for
the 386 PC. I just don't recall what major sub-systems he has in his
OS. I doubt he bothered with a file system, and I don't know if he
went as far as virtual memory or not.
Also, there was a series in Dr. Dobbs detailing the BSD386 porting
effort that might give you some information.
Finally, the only REAL cite I can give you because I mentioned to a
friend in email:
uC/OS
The Real-Time Kernel
"uC/OS is a portable, ROMable, preemptive, real-time, multitasking
kernel for microprocessors. Written in C with a touch of assembly. 266
pages w/disk".
$49.95 ISBN 0-87930-444-8
R&D Books 800-444-4881
I haven't seen this book, this was all gleaned from an ad in a UNIX
magazine.
It all really depends on what part of an OS you're interested in.
Everybody has different view of what an OS IS! Some folks think it's
just the command processor, others the file system, etc. How does
MS-DOS differ from a bunch of library calls and a command processor
and program loader? Clearly UNIX or VMS offer far more than that, but
does that make MS-DOS NOT an operating system? I dunno.
dave porter
as far as I remember, GEORGE was a retrofitted acronym.
it was called GEORGE after some George or other related
to the project, and then they had to figure out what
it stood for.
mind you, we believed it really meant
General
Environment
Organized for the
Random
Generation of
Errors.
i have used KDF9 and it didn't run GEORGE. the
machine came with a 'director' (like the 1900 'exec').
English Electric supplied "operating system" software
called, I think, PROMPT/POST. at Leeds U. we ran
a homebrew system called ELDON2. Ran about 32 program
development terminals in 32K (x 48-bit words) of memory.
GEORGE on ICL was definitely an operating system. I suppose
there could have been a language called GEORGE on KDF9.
dave
--
For email, please remove the 'w' from my address. Sorry.
David Green <dgr...@iinet.net.au> wrote in article
<352f738b...@news.m.iinet.net.au>...
> On 10 Apr 1998 22:53:09 GMT, "dave porter" <por...@wultranet.com>
> wrote:
>
> >GEORGE was the O/S for the ICL (formerly ICT) 1900
> >series of machines. I don't *think* it has anything
> >to do with Elliot.
> >
>
> My understanding is that GEORGE dates back to 1957; stood for GEneral
> ORder GEnerator; and was developed on an English Electric Deuce
> computer at the University of New South Wales in Sydney, Australia
> by Dr Charles Hamblin.
>
> It was later adopted by English Electric for its second generation
> computer called KDF9. English Electric later merged with ICT giving
Richard M. Alderson III
I'd forgotten all about that nomenclature. I was awfuly surprised, as a young
systems programmer, to learn that the "Tops-20 Operating System" class taught
by DEC covered none of the internals I was interested in (that was the "Tops-20
Monitor Internals" class), but covered the batch system and the spooled device
handlers!
Heinz W. Wiggeshoff
Alistair J. R. Young (avatar...@arkane.demon.co.uk) writes:
>
> Along similar lines, could anyone recommend a good starting point for
> someone interested in creating an operating system from more-or-less
> scratch (other than reading through, say, the Linux kernel source,
> etc., which I've already thought of)?
In the early 1970's I was peripherally involved with OS/360 performance
evaluation by simulation. The more I thought about it, the best way of
getting nitty gritty results seemed to involve re-creating OS/360!
So pick up some dusty tomes on OS/360 MFT, for example, and try to run
a program which interprets batch job commands (aka JCL). Embellish
this effort with extensions that cause small demo programs to run.
Finish the project with interactive sessions (aka TSO) running on
a microprocessor such as the Motorola 6800.
Jack Peacock
>Alistair J. R. Young (avatar...@arkane.demon.co.uk) writes:
>
> Along similar lines, could anyone recommend a good starting
point for
> someone interested in creating an operating system from
more-or-less
> scratch (other than reading through, say, the Linux kernel
source,
> etc., which I've already thought of)?
>
required to hold the manuals) OS called VMS. You can obtain the
books from DEC or I understand many are online too. The system
calls are well defined for 32 and 64 bit operations and the OS
runs on more than one hardware platform. I suggest you look at
implementing VMS on the X86 platform. It hasn't been done yet
and I'm sure it would have some commercial value too. There are
over 400,000 VMS sites in operation today, all eager to switch
to X86 hardware and escape the high prices DEC charges for
support and hardware.
Jack "of course I'm serious" Peacock
Kevin McQuiggin
For this subject a good text is "Operating Systems: Design and
Implmentation" by Andrew Tanenbaum.
Kevin
Alistair J. R. Young wrote:
>
> On Mon, 13 Apr 1998 21:28:21 +0800, in message <353212...@pc.jaring.my>,
> Zulkiflii Hamzah <zu...@pc.jaring.my> (== zulh)
> praised Shub-Internet thus:
>
> > Hello there,
> > Can somebody recommend / advice me about
> > a good starting line to write a compiler?
> > Could be book title, or... maybe some pointer?
>
> Along similar lines, could anyone recommend a good starting point for
> someone interested in creating an operating system from more-or-less
> scratch (other than reading through, say, the Linux kernel source,
> etc., which I've already thought of)?
>
> Perhaps something along the lines of "Operating System Design for
> Dummies^WReasonably Clueful People Who However Have Only Previously
> Written Application-Type Stuff"?
>
Paul Repacholi ( prep )
You may also want to look at C.mmp and Hydra, for a OS that is
different from anything else out there. Exept possibly AS400. Any
one klnow how much of the S/38 stuff has surived the transition?
The RSX sources are a good read too. Much smaller than VMS, but no
VM of course.
--
~paul ( prep ) Paul Repacholi,
1 Crescent Rd.,
erepa...@cc.curtin.edu.au Kalamunda,
+61 (08) 9257-1001 Western Australia. 6076
Megan
alde...@netcom.netcom.com (Richard M. Alderson III) writes:
>I'd forgotten all about that nomenclature. I was awfuly surprised, as a young
>systems programmer, to learn that the "Tops-20 Operating System" class taught
>by DEC covered none of the internals I was interested in (that was the "Tops-20
>Monitor Internals" class), but covered the batch system and the spooled device
>handlers!
Back when I was taking a class in 'First Line Software' (read OS) at
WPI, the description of a monitor was something which handled one
system resource, and that an OS was a collection of monitors.
Yes, a system would be pretty useless without some CUSPS, but we
never considered them as part of the OS... they were simply user
applications which we provided... system programmers were free to
write their own... and some did (a friend of mine at WPI wrote
the program 'ptys' to replace opser on the 5.07 KA system we
had, and on the -11s we had, someone wrote an out-of-address-space
debugger, called fddt, for use under V2C RT-11 on an 11/40)
Megan Gentry
Former RT-11 Developer
+--------------------------------+-------------------------------------+
| Megan Gentry, EMT/B, PP-ASEL | tcp/ip (work): gen...@zk3.dec.com |
| Unix Support Engineering Group | or: gen...@rusure.enet.dec.com |
| Digital Equipment Corporation | (non-work): m...@world.std.com |
| 110 Spitbrook Rd. ZK03-2/T43 | URL http://world.std.com/~mbg/ |
| Nashua, NH 03062 | "pdp-11 programmer, some assembler |
| (603) 884 1055 | required" - mbg |
+--------------------------------+-------------------------------------+
Thomas G. McWilliams
Alistair J. R. Young <avatar...@arkane.demon.co.uk> wrote:
: Along similar lines, could anyone recommend a good starting point for
: someone interested in creating an operating system from more-or-less
: Design for Dummies" . . .
As a first book on the subject I would recommend Douglas Comer's
"Operating System Design, the XINU Approach" ISBN 0-13-637539-1.
The target processor which is used in the book is the LSI-11. The LSI-11
is somewhat similar to CP/M class processors in architecture, memory
requirements, and processing power. The book shows that a non-trivial
multi-tasking OS can indeed run in less than 64 Kilobytes. Most of the
code is written in C so that with the proper compiler it could be made
to run on an 8080 or Z80 computer.
There is another version of Comer's book which targets the IBM PC type
computer, but it "cheats" to some degree by relying on the underlying PC
BIOS support. The LSI-11 version of the book is closer to the "iron",
and IMHO makes for a better introduction.
Here are three URLs for XINU related material:
http://www.cs.purdue.edu/homes/dec/osbooks.html
http://www.cs.purdue.edu:80/research/xinu.html
http://willow.canberra.edu.au/~chrisc/xinu.html
Andrew Tanenbaum is another author who has written several good books on
OS design but they are much broader in scope and cover theory more
extensively. Because Comer is more focused and doesn't emphasize as much
theory, the Comer book is more digestible as an introductory text.
Robert Billing
In article <353212...@pc.jaring.my>
zulh_n...@pc.jaring.my "Zulkiflii Hamzah" writes:
> Can somebody recommend / advice me about
> a good starting line to write a compiler?
This reminds me of my misspent youth...
One upon a time I was a young, hackish, and wildly entheusiastic, and
I wanted to learn compiler writing. I bought a copy of Gries "Compiler
Construction" in Foyle's bookshop, and didn't even wait to get it home,
I started reading on the train at Waterloo, and was several chapters in
by the time I got to Farnborough.
It got me nowhere.
However a few years later I managed to get on Martin Richards'
compiler writing course. Suddenly light dawned, and over about the
first four lectures I went from wondering if I would ever understand
compiler writing, to wondering why I hadn't understood it before.
Since that time a book has been published which contains most of the
material on that course. It basically is an introduction to BCPL, a
precursor of C, and a complete line by line dissection of the compiler.
The book is "BCPL The Language and the Compiler" by Richards and
Whitby-Strevens.
Colin Tan
timo...@cyberramp.net wrote:
: A pretty good book that describes these tools is "lex*yacc" published
: by O'Reilly & Associates; ISBN 1-56592-000-7.
This is an excellent recommendation. It had me up and writing my compiler (ok.. actually
it was a pre-processor that translated mathematical equations written in a self-designed
language called "MathPak" to C code) in one day.
There's also an excellent example in the book on translating SQL code to C, and another
for creating menus.
Colin Tan
Alistair J. R. Young <avatar...@arkane.demon.co.uk> wrote:
: Along similar lines, could anyone recommend a good starting point for
: someone interested in creating an operating system from more-or-less
: scratch (other than reading through, say, the Linux kernel source,
: etc., which I've already thought of)?
Something similiar to this but a lot more pleasant is getting hold of the book
"Linux Kernel Internals" by O'Reilly (I think). IMO its an
excellent place to learn how to implement kernels.
David Wragg
sa...@engin.umich.edu (Sarr J. Blumson) writes:
> In article <353212...@pc.jaring.my>,
> Zulkiflii Hamzah <zulh_n...@pc.jaring.my> wrote:
> >Hello there,
> >
> >a good starting line to write a compiler?
>
That's the old dragon book. I doubt it's available anymore.
It was revised as the new dragon book - "Compilers: Principles,
Techniques and Tools" by Aho, Sethi and Ullman. Published by
Addison-Wesley.
--
Dave Wragg
jmfb...@ma.ultranet.com
In article <ALDERSON.98...@netcom.netcom.com>,
alde...@netcom.netcom.com (Richard M. Alderson III) wrote:
>In article <6gqco0$mtc$1...@strato.ultra.net> jmfb...@ma.ultranet.com
writes:
>
>>In article <ALDERSON.9...@netcom.netcom.com>,
>> alde...@netcom.netcom.com (Richard M. Alderson III) wrote:
>
>>>I can't answer that question, but I will point out that the Tops-10 and
>>>Tops-20 operating systems have always been called "monitors";
>>>I'm not certain
>>>of the terminology for ITS in this regard.
>
>>Well, that's not quite true. The TOPS-10 and TOPS-20 operating systems
>>included the monitor and the user mode programs. Without those user mode
>>programs, timesharing wouldn't be quite as interesting or useful :-).
>>Sometimes, our use of the term operating system would also include the
>>hardware it ran on. It really depended on the context of the
conversation.
>
>as a young systems programmer, to learn that the "Tops-20 Operating
>System" class taught by DEC covered none of the internals I was
>interested in (that was the "Tops-20 Monitor Internals" class),
>but covered the batch system and the spooled device handlers!
Yea. We never could get everybody consistent w.r.t. terminology.
Product managers had a set of definitions all of their own and
one had to do an almost complete translation in order to
communicate with one of them. :-)
/BAH
jmfb...@ma.ultranet.com
In article <ErDs5...@world.std.com>, m...@world.std.com (Megan) wrote:
>
>alde...@netcom.netcom.com (Richard M. Alderson III) writes:
>
>
>>I'd forgotten all about that nomenclature. I was awfuly
>>surprised, as a young systems programmer, to learn that
>>the "Tops-20 Operating System" class taught
>>by DEC covered none of the internals I was interested in
>>(that was the "Tops-20
>>Monitor Internals" class), but covered the batch system and
>>the spooled device handlers!
>
>WPI, the description of a monitor was something which handled one
>system resource, and that an OS was a collection of monitors.
<snip>
Yes. Well, that was just plain wrong w.r.t. our mainframe OSs.
For TOPS-10 there was just one monitor :-).
Maybe that flavor of teaching was the beginning of calling
[what I think of] terminals as monitors.
/BAH
John Wilson
In article <6gvhst$dan$1...@ligarius.ultra.net>,
<jmfb...@ma.ultranet.com> wrote:
>Yea. We never could get everybody consistent w.r.t. terminology.
>Product managers had a set of definitions all of their own and
>one had to do an almost complete translation in order to
>communicate with one of them. :-)
I got the impression there was some kind of weird one-upsmanship going on
with the names of things, where everyone wants to think of a new name for
the same old stuff. In the old days there were "monitors", then later
some of them were called "executives", and of course nowadays everything
is a "kernel" even though half the world doesn't even know how to spell it
(certainly no one at Commodore did)! You just don't seem to hear words like
"sub-tasking monitor" or "hypervisor" any more...
Then a user program context might be called a "job", or a "fork", or a
"procedure", or a "task", or a "process". Or even a "ground" to the DG RDOS
folks. And somehow people think the newer names are flashier. Threads have
been around forever but now that everyone calls them that, they're suddenly
sexy. Meanwhile, if you referred to Linux as a "timesharing monitor",
what few Linux-heads knew what that meant might even take offense!
John Wilson
D Bit
Robert Hansen
Still another reference is "Writing Compilers & Interpreters" by
Ronald Mak. It is practically a tutorial in writing a Pascal
Compiler in "C". UGH!!
Bob Hansen
Kevin McQuiggin wrote:
>
> Another classic reference is "Compiler Construction for Digital
> Computers" by David Gries.
>
> Zulkiflii Hamzah wrote:
> >
> > Hello there,
> >
> > Can somebody recommend / advice me about
> > a good starting line to write a compiler?
> > Could be book title, or... maybe some pointer?
> >
John Ahlstrom
I am just now reading:
Soltis, Frank
Inside the AS/400:Featuring the AS/400e
(second edition)
available from amazon.com
This is in the same league (but I think it couldn't
make the playoffs) as Bucholz on Stretch, Thornton on CDC6600
and Wulf on Hydra.
According to it all the System/38 stuff survives in the AS/400.
John Ahlstrom
jahl...@cisco.com
Using Java to Decrease Entropy
--
Do something small, useful now.
Dave Fuggle
Robert Billing wrote:
[snip]
> It then goes on to describe the minimum configuration:
>
> 48K words of core[1]
> 2M words of direct access backing store
>
> [1] This would have been 48K x 36 bits or just over 200K bytes
>
The two 1904's that I used when I were a lad were described to me as
having 24bit words (Indeed IIRC the 1903 [or was it a 1902? is it still
there?] at the Science Mueseum was similarly described...)
Dave
Charlie Gibbs
In article <892538...@tnglwood.demon.co.uk>
uncl...@tnglwood.demon.co.uk (Robert Billing) writes:
> Since that time a book has been published which contains most of the
>material on that course. It basically is an introduction to BCPL, a
>precursor of C, and a complete line by line dissection of the compiler.
>
> The book is "BCPL The Language and the Compiler" by Richards and
>Whitby-Strevens.
I have a copy of a similar book dealing with the small-C compiler
(written for CP/M). I think it's just called "The Small-C Compiler"
and is written by James E. Hendrix (probably not the same guy who
used to set guitars on fire). It contained the entire source code
for the compiler, with a detailed analysis of how it worked.
Although small-C wasn't a complete C compiler, it implemented a
pretty good subset for something that could fit into a 64K CP/M
box.
I don't know whether the book is available anymore, but if I find
it again I could post the ISBN if anyone is interested.
--
cgi...@sky.bus.com (Charlie Gibbs)
Remove the first period after the "at" sign to reply.
Alan Bowler
In article <35336F8F...@zianet.com> Robert Hansen <rlha...@zianet.com> writes:
>Still another reference is "Writing Compilers & Interpreters" by
>Ronald Mak. It is practically a tutorial in writing a Pascal
>Compiler in "C". UGH!!
You wouldn't want to write it in Pascal. :-(
Richard M. Alderson III
In article <6gu9j8$a27$1...@news.skylink.net> "Jack Peacock" <pea...@simconv.com>
writes:
>I suggest you look at implementing VMS on the X86 platform. It hasn't been
>done yet and I'm sure it would have some commercial value too.
Dave Cutler might disagree, but isn't Windows NT the X86 implementation of VMS?
Robert Schuldenfrei
Amen Brother Wilson :) I was shocked, SHOCKED, to find out that conjoint
analysis was multiple regression. Not to mention that regression is really
just correlation which is just an application of the least squares method of
what you calculus types might call fitting at line so as to minimize the
distance to a set of data points. I rest my case.
Cheers,
Bob
John Wilson wrote in message <35337...@news.wizvax.net>...
Jim Maurer
While you can't read the manuals on the web, you can order them. I bought a
bunch of old S/360 and 370 manuals a little while back from IBM's web pages.
"IBM System/360 Operating System Introduction" is order number GC28-6534-4.
Most of those old manuals are now available for little more than the
shipping cost.
Robert Schuldenfrei wrote in message <6gqhib$c...@news-central.tiac.net>...
>It is interesting to note as an American, how much really good and novel
>ideas came from the UK. The works of Turing and Wilkes not withstanding,
>who are the folks who came up with George? I consider myself somewhat a
>student of computer history, but I guess I should make that American
>computer history :)
>
>As I stated earlier, I really do not know what IBSYS could do, let alone
>George. Could someone post a feature listing of both systems? I once had
>an introduction to OS/360 that was printed BEFORE it was released. It has
>long since been canned. It would be instructive to also list the features
>of it, the actual features of DOS (the mainframe "stop gap" system that was
>released as a substitute for OS/360), and then OS/360 as released. It
would
>be really great if IBM published on the web, those early manuals and
>brochures.
Richard M. Alderson III
In article <6h0mjb$kld$1...@nntp2.uunet.ca> atbo...@thinkage.on.ca (Alan Bowler)
writes:
Funny, the Pascal compilers with which I'm familiar were written in Pascal:
Naegel's DEC-10 compiler (originally cross-compiled on the 6600 at ETH), which
was adopted for the 370 architecture on the Hitachi 8000 by Hitachi, then
translated to English by some folks at the Australian Atomic Energy Commission.
The documentation for the last introduced me to Backus-Naur Forms (which I
immediately recognized as some kind of context-free phrase structure grammar,
and used to learn the language really well).
Then there was Larry XXX's Pascal/VS compiler for VM/CMS and OS/MVS, written at
IBM's Santa Teresa Labs. Different extensions to the language than devised by
the Hitachi/AAEC folks.
The DEC-10 compiler was revised and made Tops-20 native by Chuck Hedrick at
Rutgers. Hundreds of Stanford students learned to program in Pascal at LOTS,
using this one.
Well, OK, there's the tiny-Pascal compiler written in NorthStar BASIC and
published in BYTE, Sep.-Nov. 1978 (or was it Oct.-Dec.?), with a small runtime
written in Z80 machine language. This was my introduction to Pascal; a friend
at the University of Chicago Comp Center, Paul Collard, wanted to port it to
Tops-20 (we didn't know about the Naegel Tops-10 compiler), so we started
working on it, typing in the source to BASIC+2. I decided we needed more
information on how to do a compiler, so I started reading the green dragon book
(Aho & Ullman--it was brand new at the math library), and started translating
the BASIC into Pascal for a bootstrap. (I also got a copy of the Tops-10
Monitor Calls manual--we didn't *have* a Tops-20 manual in the applications
group--to try to set up the runtime system. Good thing I didn't get too far.)
We got the BASIC version to work (mostly), but before we could start the next
phase, inputting my Pascal code for the Pascal compiler, two things happened:
Tim Little got a copy of the Tops-20 compiler from Joe Green at 3M, and Dave
Farb got a copy of the AAEC compiler for the Amdahl 470, and so we sort of lost
interest. But were it not for this effort, I'd have finished my Ph. D. in
Indo-European linguistics and never have gotten to do all the fun stuff I've
done for the last 20 years.
Robert Billing
In article <35335804...@vocalis.com>
david....@vocalis.com "Dave Fuggle" writes:
> The two 1904's that I used when I were a lad were described to me as
> having 24bit words (Indeed IIRC the 1903 [or was it a 1902? is it still
> there?] at the Science Mueseum was similarly described...)
Now you've got me confused, and I'm relying on 25 year old memories.
At the time I was using Honeywell G265 (18 bits) IBM1130 (16 bits)
IBM370 (32 Bits) and an early PDP11 (16 Bits). I'm sure the 1900 series
were a multiple of 6, and used a 6 bit character set, but 24v36?
Unfortunately the person who could have given us a definitive answer
(my father) died about 18 months ago, but I have a lot of his notes on
1900 series, and I may be able to give you a better answer.
What worries me about 24 bits being the word length is that it is very
inconvenient for the sort of floating point we used to use in those
days. 24 is too short to get single precision in 1 word, and going to 2
words for *all* fp calculations would have been too slow to be
acceptable. The C idea of going straight to 64 bits all the time just
isn't viable with the sort of fp hardware that takes a shocking number
of cycles to do double precision.
Jo Meder
"Charlie Gibbs" <cgi...@sky.bus.com> writes:
> I have a copy of a similar book dealing with the small-C compiler
> (written for CP/M). I think it's just called "The Small-C Compiler"
> and is written by James E. Hendrix (probably not the same guy who
> used to set guitars on fire). It contained the entire source code
> for the compiler, with a detailed analysis of how it worked.
"A Small C Compiler"; James E. Hendrix; M & T Books; Redwood City, CA 1988
ISBN 0-934375-88-7
ISBN 1-55851-007-9 (disk)
ISBN 0-934375-97-6 (book & disk)
HTH.
Regards,
Jo
--
j...@delorges.in-berlin.de --- Berlin, Germany
http://delorges.in-berlin.de/
Gene Wirchenko
atbo...@thinkage.on.ca (Alan Bowler) wrote:
>In article <35336F8F...@zianet.com> Robert Hansen <rlha...@zianet.com> writes:
>>Still another reference is "Writing Compilers & Interpreters" by
>>Ronald Mak. It is practically a tutorial in writing a Pascal
>>Compiler in "C". UGH!!
>
>You wouldn't want to write it in Pascal. :-(
You wouldn't want to write it period.
See
http://www.lysator.liu.se/c/bwk-on-pascal.html
which is Brian W. Kernighan's "Why Pascal Is Not My Favorite
Programming Language". The article is a hoot. It's also tragic that
such a pathetic excuse for a language is actually Out [Th|H]ere.
Sincerely,
Gene Wirchenko
C Pronunciation Guide:
y=x++; "wye equals ex plus plus semicolon"
x=x++; "ex equals ex doublecross semicolon"
Colin Tan
Charlie Gibbs <cgi...@sky.bus.com> wrote:
: I have a copy of a similar book dealing with the small-C compiler
: (written for CP/M). I think it's just called "The Small-C Compiler"
: and is written by James E. Hendrix (probably not the same guy who
: used to set guitars on fire). It contained the entire source code
: for the compiler, with a detailed analysis of how it worked.
I remember one book by Niklaus Wirth (i.e. Father of Pascal) on
writing a compiler. He takes you step by step thru writing
a real compiler with source code and all. I forget what
its called, but I'm sure someone here will know the
title.
Hung Michael Nguyen
In article <6gvf2c$71...@id4.nus.edu.sg>,
Along the same lines, "The Design and Implementation of the 4.4 BSD Operating
System" by McKusick, Bostic, et al is a very good look at the 4.4 BSD kernel.
You can then experiment using {Free|Net|Open}BSD. If you are into SysVR4,
"The Magic Garden Explained" (I can't remember the authors) is a good look
at the SysV Rel. 4 kernel.
However, all three books are less useful when it comes to writing a OS from
scratch. For that, you probably want to bite off something smaller. The
Xinu book by Comer is pretty good, as is the book by Tanenbaum, "Operating
Systems: Design and Implementation", which I think has been recently revised.
Relatedly, "Computer Organization and Design: The Hardware/Software Interface"
by Hennessy and Patterson is also a good read for anyone interested in OSes.
Mike.
--
<HTML><META HTTP-EQUIV="Content-Type:text/html"> <SCRIPT>
function X() {var Text = "HTML is not acceptable for using in mail " +
"or usenet so your browser will stop."; alert(Text); parent.close();};
</SCRIPT> </HEAD><BODY onLoad="X();return true">Hi</HTML>
David Green
On 11 Apr 1998 11:09:51 GMT, and...@cucumber.demon.co.uk (Andrew
Gabriel) wrote:
>Relative geneology of ICT/ICL and Elloitt are shown at
>
> http://www.cucumber.co.uk/geccl/19471972/index.html
>
Can we please check this url? I would like to look at the web site
but my server returns the following:
While trying to retrieve the URL:
http://www.cucumber.co.uk/geccl/19471972/index.html
The following error was encountered:
Unable to determine IP address from host name for
www.cucumber.co.uk
The dnsserver returned:
DNS Domain 'www.cucumber.co.uk' is invalid: Host not found
(authoritative).
Andrew Gabriel
In article <353450b4...@news.m.iinet.net.au>,
dgr...@iinet.net.au (David Green) writes:
>On 11 Apr 1998 11:09:51 GMT, and...@cucumber.demon.co.uk (Andrew
>Gabriel) wrote:
>
>>Relative geneology of ICT/ICL and Elloitt are shown at
>>
>> http://www.cucumber.co.uk/geccl/19471972/index.html
>>
>
>Can we please check this url?
Sorry, I screwed up. :-(
Try: http://www.cucumber.demon.co.uk/geccl/19471972/index.html
--
Andrew Gabriel
Consultant Software Engineer
Richard M. Alderson III
In article <6h1cqn$9b...@id4.nus.edu.sg> Colin Tan
<tank...@luakt-r2.iscs.nus.edu.sg> writes:
>I remember one book by Niklaus Wirth (i.e. Father of Pascal) on writing a
>compiler. He takes you step by step thru writing a real compiler with source
>code and all. I forget what its called, but I'm sure someone here will know
>the title.
Well, there's the chapter on "PL/0" (a Pascal subset) in his _Algorithms + Data
Structures = Programs_, in which he discusses parsing and lexical scanners and
the like. Very early dialect of Pascal in the examples in the book.
Then there's his Modula-2 book, but it's not really a step-by-step on compiler
writing...
David Green
On Sun, 12 Apr 98 08:46:29 GMT, Robert Billing
<uncl...@tnglwood.demon.co.uk> wrote:
> dgr...@iinet.net.au "David Green" writes:
>> .... I think it was probably a programming language rather
>> than an operating system, and that there were both interpreters and
>> compilers for it.
>
> I have a George manual, ICL Part number 4345(1.76). It is a loose
>collection of pages about one inch thick. I can't find a date on it but
>the 76 may be the year. The paper size is A4, which fixes it to be
>after the mid 60s, which is when the UK went over from the old imperial
>paper sizes to "A" sizes.
>
> The preface begins:
> "GEORGE 3 and 4 are the operating systems for the larger computers of
>the 1900 series."
>
> It then goes on to describe the minimum configuration:
> 48K words of core[1]
> 2M words of direct access backing store
>
>[1] This would have been 48K x 36 bits or just over 200K bytes
>
Umm.. Couldn't be any clearer than that.
On the other hand, my source (see below) says " GEORGE, Alphacode and
GIP were popular languages at UTECOM".
The KDF9 and the 1900 series were contempory machines. I cut some code
for the KDF9 in 1963, but there was no KDF9 to test it on. Sydney
University took delivery of one of the first KDF9's in (I think) 1964.
I have an ICT 1900 Series coding summary card dated "Date of
Publication - March 1965". It mentions the 1902, 1903, 1904, 1906 and
1907. Mount Isa Mines in Queensland took delivery of a 1909 in March
1966. (Incidently, that 1909 had a 16k 24 bit word memory). And I must
admit I can't remember coming across GEORGE on either the KDF9 or the
1909.
So, is it just a coincidence that George 1 and George 2 were developed
on Deuce and ported to the English Electric KDF9, and that George 3
and George 4 were o/s for the 1900 series? Or did the 1950's George
language metamorphose into the 1960's George o/s?
Is this, perhaps, an example of how an early o/s evolved?
Regards
David Green
ps 1: My "source" is "UTECOM - An English Electric DEUCE" by
R. A. Vowels of the Dept of Computer Science, Royal Melbourne
Institute of Technology, 1993. To quote:
"GEORGE (GEneral ORder GEnerator) was an algebraic language
characterized by the suffix notation developed by the logician
Dr Charles Hamblin ... Mathematical notation transformed naturally to
GEORGE language constructs, particularly in the representation of
vector and matrix operations. The GEORGE language provided for the
evaluation of expressions given in the somewhat unusual reverse Polish
form. It included conditionals, loop control, jumps, subroutines,
array input and output, and built-in functions.
"Several versions of the interpreter were written for program testing,
namely GEORGE IB, GEORGE II and GEORGE IIC along with companion
compilers. GEORGE II was available in 1958.
"Hamblin's language and GEORGE were pioneering achievements in
computer development, for the logic design was taken up by The English
Electric Company for its KDF9 computer ...
"In 1964, GEORGE continued to be used for research and contract work,
as well as for undergraduate teaching."
ps 2: UTECOM was an acronym of the capital letters in University of
Technology Electronic COMputer.
Martin Taylor
Robert Billing wrote:
>
> In article <35335804...@vocalis.com>
> david....@vocalis.com "Dave Fuggle" writes:
>
> > The two 1904's that I used when I were a lad were described to me as
> > having 24bit words (Indeed IIRC the 1903 [or was it a 1902? is it still
> > there?] at the Science Mueseum was similarly described...)
>
> Now you've got me confused, and I'm relying on 25 year old memories.
> At the time I was using Honeywell G265 (18 bits) IBM1130 (16 bits)
> IBM370 (32 Bits) and an early PDP11 (16 Bits). I'm sure the 1900 series
> were a multiple of 6, and used a 6 bit character set, but 24v36?
>
> Unfortunately the person who could have given us a definitive answer
> (my father) died about 18 months ago, but I have a lot of his notes on
> 1900 series, and I may be able to give you a better answer.
>
> What worries me about 24 bits being the word length is that it is very
> inconvenient for the sort of floating point we used to use in those
> days. 24 is too short to get single precision in 1 word, and going to 2
> words for *all* fp calculations would have been too slow to be
> acceptable. The C idea of going straight to 64 bits all the time just
> isn't viable with the sort of fp hardware that takes a shocking number
> of cycles to do double precision.
The 1900 was a 24-bit machine. Programs used either 15-bit direct
addressing, or 22-bit relative or replaced addressing. Integer range was
-2**23 to 2**23-1. Personally I never had anything to do with
floating-point, being a COBOL programmer, but 48 bits certainly rings a
bell. I have a feeling that floating point was implemented at least
partly in software on most medium-range 1900s: was it the 1907 and 1909
among early models that included FP hardware? Possibly the later "S" and
"T" models did as well.
The thing in the Science Museum was a 1903A, I seem to recall. We were
still using one where I worked, at the time they acquired it :-).
Martin Taylor
jmfb...@ma.ultranet.com
In article <6gvi3b$dan$2...@ligarius.ultra.net>,
I occurred to me last night that the thingies that Megan described
as monitors were actually called modules. Just thought I'd
add more arcane nomenclature in the mix :-).
/BAH
David Spencer
On Tue, 14 Apr 98 21:49:15 GMT, uncl...@tnglwood.demon.co.uk (Robert
Billing) wrote, net of snippage:
> Now you've got me confused, and I'm relying on 25 year old memories.
>At the time I was using Honeywell G265 (18 bits) IBM1130 (16 bits)
>IBM370 (32 Bits) and an early PDP11 (16 Bits). I'm sure the 1900 series
>were a multiple of 6, and used a 6 bit character set, but 24v36?
>
> Unfortunately the person who could have given us a definitive answer
>(my father) died about 18 months ago, but I have a lot of his notes on
>1900 series, and I may be able to give you a better answer.
>
> What worries me about 24 bits being the word length is that it is very
>inconvenient for the sort of floating point we used to use in those
>days. 24 is too short to get single precision in 1 word, and going to 2
>words for *all* fp calculations would have been too slow to be
>acceptable. The C idea of going straight to 64 bits all the time just
>isn't viable with the sort of fp hardware that takes a shocking number
>of cycles to do double precision.
The 1900 Series was 24 bit. Floating point data types were 48 bits
single precision, 96 bits double precision. The '2M words of direct
access backing store' you referred to earlier was basically a swap file.
It has to be said that 48k was *hopeless* for George 3. Typically only
two or three realistic user jobs would fit simultaneously in core even
with 96k. This was one of the driving factors behind the invention of
paged memory...
--
David Spencer
Logica BT OMC Team, Ipswich
These are personal opinions, not shared by Logica or BT,
and worth exactly what you paid for them.
Robert Hansen
Alan Bowler wrote:
>
> In article <35336F8F...@zianet.com> Robert Hansen <rlha...@zianet.com> writes:
> >Still another reference is "Writing Compilers & Interpreters" by
> >Ronald Mak. It is practically a tutorial in writing a Pascal
> >Compiler in "C". UGH!!
>
> You wouldn't want to write it in Pascal. :-(
I sure would! :-)
Alan Bowler
In article <ALDERSON.98...@netcom.netcom.com> alde...@netcom.com writes:
>In article <6h0mjb$kld$1...@nntp2.uunet.ca> atbo...@thinkage.on.ca (Alan Bowler)
>
>>>Still another reference is "Writing Compilers & Interpreters" by Ronald Mak.
>>>It is practically a tutorial in writing a Pascal Compiler in "C". UGH!!
>
>>You wouldn't want to write it in Pascal. :-(
>
You missed the ":-(".
However, there was something behind my comment. I seriously doubt that
the compilers you are familiar with were written in Pascal as defined
by the Standard. It is more likely that they were written in some
extended "Pascal like language". In particular, standard Pascal does
not allow any form of separate compilation, and the idea of trying to
maintain a whole compiler as a single huge module module makes me
shudder.
Baz
A book which I really liked is "Writing Interactive Compilers And
Interpreters" by PJ BROWN ISBN 0 471 27609 X (published by Wiley). Not sure
if its still in print, but a wonderfully well written, and very readable
computer book. Rare indeed.
It discusses many important issues including the use of automated tools,
design decisions, with many easily overlooked pitfalls highlighted, well
worth the reading.
................. Barry
timo...@cyberramp.net wrote in message
<6gt7vt$kks$1...@newshost.cyberramp.net>...
>Zulkiflii Hamzah <zu...@pc.jaring.my> wrote:
>
>>Hello there,
>
>>Can somebody recommend / advice me about
>>a good starting line to write a compiler?
>>Could be book title, or... maybe some pointer?
>
>>fyi I have developed several applications in assembly language.
>>and, fairly OK-to-competitive in C programming.
>
>
>HI. If you're wanting to write a compiler f4rom scratch, and how
>little or no experience, I would suggest that you consider going the
>YACC/LEX route. This will still not reduce the problem to trivial
>proportions, but will help. You will have to create a "syntax" for
>yacc, then compile it.
>
>YACC and LEX are compiler writing toos that come to us from the unix
>world. they are both still very popular in the unix comunity, but
>therre are versions available for the MS-DOS world.
>
>A pretty good book that describes these tools is "lex*yacc" published
>by O'Reilly & Associates; ISBN 1-56592-000-7.
>
>Probably the biggest disadvantage to this approach is that it results
>in a compiler that is usualy far from optimal.
>
>I would also suggest that you check out the DJGPP newsgroup. I don't
>have the complete usenet address handy, but you can search for DJGPP
>and find it pretty fast. There are a number of people on that ng who
>write compilers, and are willing to help. You canm also get the source
>to the GNU/DJGPP compilers and play with them.
>
>Good luck.
>
>
>
>Tim Olmstead
>webmaster of the CP/M Unofficial web page
>email : timo...@cyberramp.net
>MAIN SITE AT : http://cdl.uta.edu/cpm
>MIRROR AT : http://www.mathcs.emory.edu/~cfs/cpm
>
Dr. Peter Kittel
In article <01bd64d3$578d0a60$0ba17392@glastonbury> "dave porter" <por...@wultranet.com> writes:
>
>GEORGE was the O/S for the ICL (formerly ICT) 1900
>series of machines.
Yeah, GEORGE! Our uni in Braunschweig, Germany, used it for years
on our ICL (indeed formerly ICT) mainframe. And under it, we used
the ALFONS command to run our Algol 60 programs.
And the boss of the computer center had Georg as first name, so
everyone thought the OS name were a pun for his one :-).
--
Best Regards, Dr. Peter Kittel // http://www.pios.de of PIOS
Private Site in Frankfurt, Germany \X/ office: peterk @ pios.de
Joseph J. Ambrose
Heinz W. Wiggeshoff wrote in message
<6gtlv2$k...@freenet-news.carleton.ca>...
>
>Kevin McQuiggin (mcqu...@sfu.ca) writes:
>> Another classic reference is "Compiler Construction for Digital
>> Computers" by David Gries.
>
> Even when new, my copy was hard on the eyes. It was printed from
> a photoreduced printout made with a 1403 running a TN chain.
So that explains the headaches I was getting from the compiler class I took
in undergrad....... :-)
Thanks Bob Sebesta...... !
Joseph Ambrose
NT Network Administrator /
Open VMS System Manager
The Conference Board
amb...@conference-board.org
Shez
s...@dsds.demon.co.uk (David Spencer) writes:
|
| The 1900 Series was 24 bit. Floating point data types were 48 bits
| single precision, 96 bits double precision. The '2M words of direct
| access backing store' you referred to earlier was basically a swap file.
|
| It has to be said that 48k was *hopeless* for George 3. Typically only
| two or three realistic user jobs would fit simultaneously in core even
| with 96k. This was one of the driving factors behind the invention of
| paged memory...
This is where the 32 bit word 2900 series came in (c1980), with it's VME
(Virtual Machine Environment) OS. For backwards compatability 2900's
could also run the 1900's George 3, which being non-virtual was now
known as DME (Direct Machine Environment). Unfortunately ICL hit a snag
here - they couldn't get customers to upgrade to VME, as without the
virtual addressing overhead DME was much faster! However, (and most
impressively I always thought) the machines could actually run the DME &
VME OS's simultaneously, despite one being a 24 bit OS and the other a
32 bit OS, as the timesharing went right down to microcode level! So you
had two sets of microcode running at the same time, being swapped in and
out with the OS. Was this what was known as a bit-sliced processor? I
recall the term was in vogue around then.
And then there was the ME29 mini which was reckoned to outperform
top-flight mainframes due to the innovative CAFS (Contents Addressable
File Store), something which I think was later fitted to all the ICL
range. Some bright spark had realised that in commercial applications
(such as database work) masses of data are fetched from disk only to be
discarded, so rather than clog up the CPU with irrelevant data, simple
pattern-matching processors were put in the disk drive electronics. The
CPU then sent to the disk to say what keys to look for, and as a
consequence 98% of the data never got much further than the disk heads,
freeing up bus bandwidth and processor time.
-Shez.
____________________________________________________
If replying by email delete .junkblok from address
John Hughes
John Hughes <jo...@AtlanTech.COM> writes:
> George was ICT wasn't it? Considerably older than IBSYS?
^^^^^
Cretin, you mean younger.
--
John Hughes <jo...@AtlanTech.COM>,
Atlantic Technologies Inc. Tel: +33-1-43204546
24 rue Montbrun, Fax: +33-1-43204579
75014 PARIS.
John Hughes
Robert Billing <uncl...@tnglwood.demon.co.uk> writes:
> [the manual for George, an OS for ICL 1900 machines] then goes on
> to describe the minimum configuration:
>
> 48K words of core[1]
> 2M words of direct access backing store
>
> [1] This would have been 48K x 36 bits or just over 200K bytes
Nope, 48K x 24 bits, or 192k bytes. Four six-bit bytes per 24 bit
word.
Megan
jmfb...@ma.ultranet.com writes:
>In article <6gvi3b$dan$2...@ligarius.ultra.net>,
> jmfb...@ma.ultranet.com wrote:
>>Yes. Well, that was just plain wrong w.r.t. our mainframe OSs.
>>For TOPS-10 there was just one monitor :-).
>>
>>Maybe that flavor of teaching was the beginning of calling
>>[what I think of] terminals as monitors.
I never said that I agreed with it...
>I occurred to me last night that the thingies that Megan described
>as monitors were actually called modules. Just thought I'd
>add more arcane nomenclature in the mix :-).
This was weirder than that, since these 'monitors' which handled
a resource could be spread over more than one module... they
just coordinated things with mutexes...
or so the instructor said...
I'd already been doing development of user programs under RT
(along with modifications to the *monitor* files :-) for a
year or so by that time, so I was taking the course for the
credit... (1976)
Megan Gentry
Former RT-11 Developer
+--------------------------------+-------------------------------------+
| Megan Gentry, EMT/B, PP-ASEL | Internet (work): gen...@zk3.dec.com |
| Unix Support Engineering Group | (home): m...@world.std.com |
| Digital Equipment Corporation | |
| 110 Spitbrook Rd. ZK03-2/T43 | URL: http://world.std.com/~mbg/ |
| Nashua, NH 03062 | "pdp-11 programmer - some assembler |
| (603) 884 1055 | required." - mbg |
+--------------------------------+-------------------------------------+
Gerardo Ospina
Colin Tan <tank...@luakt-r2.iscs.nus.edu.sg> wrote:
: Charlie Gibbs <cgi...@sky.bus.com> wrote:
: : I have a copy of a similar book dealing with the small-C compiler
: : (written for CP/M). I think it's just called "The Small-C Compiler"
: : and is written by James E. Hendrix (probably not the same guy who
: : used to set guitars on fire). It contained the entire source code
: : for the compiler, with a detailed analysis of how it worked.
: I remember one book by Niklaus Wirth (i.e. Father of Pascal) on
: writing a compiler. He takes you step by step thru writing
: a real compiler with source code and all. I forget what
: its called, but I'm sure someone here will know the
: title.
Compiler Construction (1996) ISBN 0201403536 Addison Wesley
Colin Tan
Richard M. Alderson III <alde...@netcom6.netcom.com> wrote:
: Well, there's the chapter on "PL/0" (a Pascal subset) in his _Algorithms + Data
: Structures = Programs_, in which he discusses parsing and lexical scanners and
: the like. Very early dialect of Pascal in the examples in the book.
Well, no.. this was a whole book.
Oddly enough I had found the book in the *business* library here in my U.
jmfb...@ma.ultranet.com
In article <ErHEA...@world.std.com>, m...@world.std.com (Megan) wrote:
>jmfb...@ma.ultranet.com writes:
>
>>In article <6gvi3b$dan$2...@ligarius.ultra.net>,
>> jmfb...@ma.ultranet.com wrote:
>>>Yes. Well, that was just plain wrong w.r.t. our mainframe OSs.
>>>For TOPS-10 there was just one monitor :-).
>>>
>>>Maybe that flavor of teaching was the beginning of calling
>>>[what I think of] terminals as monitors.
>
>I never said that I agreed with it...
>
>>I occurred to me last night that the thingies that Megan described
>>as monitors were actually called modules. Just thought I'd
>>add more arcane nomenclature in the mix :-).
>
>This was weirder than that, since these 'monitors' which handled
>a resource could be spread over more than one module... they
>just coordinated things with mutexes...
Oh, barf.
>
>or so the instructor said...
Did you ever figure out where the instructor got his training?
>
>I'd already been doing development of user programs under RT
>(along with modifications to the *monitor* files :-) for a
>year or so by that time, so I was taking the course for the
>credit... (1976)
Should have been a snap as long as you had a translator :-).
/BAH
Jerry Avins
MY RETURN ADDRESS IS ALTERED TO FOIL SPAMBOTS.
******* REMOVE THE "x" TO REPLY *******
The CGG version of YACC is called BISON. I have it somewhere, and maybe
also LEX or equivalent. Both for MS-DOS. Prompt me to look if you need
it.
Jerry
--
Engineering is the art | Let's talk about what
of making what you want | you need; you may see
from things you can get. | how to do without it.
----------------------------------------------------------
Jerry Avins
MY RETURN ADDRESS IS ALTERED TO FOIL SPAMBOTS.
You don't say what language you want to compile, but C seems to be
implied. If you aren't familiar with Forth, I suggest that you look into
that before making a decision. There are several good write-your-own
Forth books, and a lot of information at www.forth.org. Many who have
written Forth compilers are active at comp.lang.forth. www.taygetta.com
also has a lot of material, as does the Forth/UK site.
Jerry
--
Engineering is the art | Let's talk about what
of making what you want | you need; you may see
from things you can get. | how to do without it.
----------------------------------------------------------
Zulkiflii Hamzah wrote:
>
> Hello there,
>
> Can somebody recommend / advice me about
> a good starting line to write a compiler?
> Could be book title, or... maybe some pointer?
>
> fyi I have developed several applications in assembly language.
> and, fairly OK-to-competitive in C programming.
>
> Thank you very much.
> Zul Hamzah.
Russell Schulz
Colin Tan <tank...@luakt-r2.iscs.nus.edu.sg> writes:
> I remember one book by Niklaus Wirth (i.e. Father of Pascal) on
> writing a compiler.
Compiler Construction
Addison-Wesley
0-201-40353-6 (disk included)
1996
This book has emerged from my lecture notes for an introductory
course in compiler design at ETH Zürich. Several times I have
been asked to justify this course, since compiler design is
considered a somewhat esoteric subject, practised only in a few
highly specialized software houses. Because nowadays everything
which does not yield immediate profits has to be justified, I
shall try to explain why I consider this subject as important and
relevant to computer science students in general.
--
Russell...@locutus.ofB.ORG Shad 86c
Alan Bowler
In article <353631...@erols.com> Jerry Avins <jxya...@erols.com> writes:
>MY RETURN ADDRESS IS ALTERED TO FOIL SPAMBOTS.
>
>The CGG version of YACC is called BISON. I have it somewhere, and maybe
>also LEX or equivalent. Both for MS-DOS. Prompt me to look if you need
>it.
>
Jerry Avins
Jerry Avins wrote:
>
> MY RETURN ADDRESS IS ALTERED TO FOIL SPAMBOTS.
> ******* REMOVE THE "x" TO REPLY *******
>
> The CGG version of YACC is called BISON. I have it somewhere, and maybe
> also LEX or equivalent. Both for MS-DOS. Prompt me to look if you need
> it.
>
> --
> Engineering is the art | Let's talk about what
> of making what you want | you need; you may see
> from things you can get. | how to do without it.
> ----------------------------------------------------------
> >
> > Zulkiflii Hamzah <zu...@pc.jaring.my> wrote:
> >
> > >Hello there,
> >
> > >Can somebody recommend / advice me about
> > >a good starting line to write a compiler?
> > >Could be book title, or... maybe some pointer?
> >
> > >fyi I have developed several applications in assembly language.
> > >and, fairly OK-to-competitive in C programming.
> >
lis...@zetnet.co.uk
On 1998-04-11 dgr...@iinet.net.au said:
:On 10 Apr 1998 22:53:09 GMT, "dave porter" <por...@wultranet.com>
:wrote:
:>GEORGE was the O/S for the ICL (formerly ICT) 1900
:>series of machines. I don't *think* it has anything
:>to do with Elliot.
:My understanding is that GEORGE dates back to 1957; stood for
:GEneral ORder GEnerator; and was developed on an English Electric
:Deuce computer at the University of New South Wales in Sydney,
:Australia by Dr Charles Hamblin.
ok, here's where we have to 'fess up. we confused the two georges. :>
however, since the debate it generated has been wondrous to behold, we
refuse to apologise, instead wondering idly if anyone has any examples
of code written in the GEORGE that was the language.
--
Communa (together) we remember... we'll see you falling
you know soft spoken changes nothing to sing within her...
lis...@zetnet.co.uk
On 1998-04-14 d...@getsetgo.doc.ic.ac.uk said:
:> >Can somebody recommend / advice me about
:> >a good starting line to write a compiler?
:> >Could be book title, or... maybe some pointer?
:> The classic book is by Aho and Ullmann, "Principles of Compiler
:>Design". You might also want to look at the newsgroup comp.
:compilers.
:That's the old dragon book. I doubt it's available anymore.
:It was revised as the new dragon book - "Compilers: Principles,
:Techniques and Tools" by Aho, Sethi and Ullman. Published by
:Addison-Wesley.
highly recommended. this is the book that enabled us to actually
understand LR(x) parsing. we still went off and wrote a recursive
descent parser for the coursework, buuut... ;>
Charles Richmond
Since this thread has degenerated into discussions of Pascal:
There is a book with *lots* of Pascal info: the title is _Pascal: The
Language and Its Implementation_, and it is edited by Baron. I do *not* have
it or any more identifying details. It contains articles by several people
who worked on developing Pascal compilers and interpreters. Niklaus Wirth's
Pascals interepreter source is included. (Be aware that the key word array is
*not* sorted properly, and the interpreter is attempting to binary search it.)
This book also has articles by a man who wrote a Pascal compiler in old-style
Pascal for the CDC6600 at ETH. This compiler has been adapted to other
machines--I used to have a paper copy for a Data General Eclipse.
For writing Forth-like interpreters, the best book is _Threaded Interpretive
Languages_, by R. G. Loeliger, Byte Books, 1981, ISBN: 0-07-039360-X. Along
with explaining how these langauges work, this book develops a TIL for the Z80 microprocessor.
These out-of-print books can be obtained through inter-library loan. Call
your local library and ask about it. It takes a few weeks to obtain the book,
but it is the *best* way I know of.
--
+-------------------------------------------------------------+
| Charles and Francis Richmond <rich...@plano.net> |
+-------------------------------------------------------------+
Paul Repacholi ( prep )
Ah, this is the `monitor' software concept af Hoar et al.
--
~paul ( prep ) Paul Repacholi,
1 Crescent Rd.,
erepa...@cc.curtin.edu.au Kalamunda,
+61 (08) 9257-1001 Western Australia. 6076
Richard M. Alderson III
In article <6h2q2p$gpu$1...@nntp3.uunet.ca> atbo...@thinkage.on.ca (Alan Bowler)
writes:
>In article <ALDERSON.98...@netcom.netcom.com> alde...@netcom.com
>writes:
>>In article <6h0mjb$kld$1...@nntp2.uunet.ca> atbo...@thinkage.on.ca
>>(Alan Bowler) writes:
>>>You wouldn't want to write it in Pascal. :-(
>You missed the ":-(".
No, I didn't, I just misunderstood it. I *thought* you were harshing on
Pascal, not commiserating with one or more weaknesses. Sorry 'bout that.
>However, there was something behind my comment. I seriously doubt that the
>compilers you are familiar with were written in Pascal as defined by the
>Standard. It is more likely that they were written in some extended "Pascal
>like language". In particular, standard Pascal does not allow any form of
>separate compilation, and the idea of trying to maintain a whole compiler as a
>single huge module module makes me shudder.
Most, if not all, of the compilers with which I am familiar do indeed include
extensions to the standard language, including "external" or similar declara-
tions for modular programming.
However, in the cases of the Dec-10/DEC-20 and Hitachi/AAEC compilers, the
whole thing was a single Pascal program, compilable in a single run. It was
an LALR(1) top-down implementation (easy enough to do in Pascal--as shown by
the "PL/0" compiler in Wirth's book).
--
Rich Alderson Last LOTS Tops-20 Systems Programmer, 1984-1991
last name @ XKL dot COM Current maintainer, MIT TECO EMACS (v. 170)
Megan
erepa...@cc.curtin.edu.au (Paul Repacholi ( prep )) writes:
>Ah, this is the `monitor' software concept af Hoar et al.
Thanks... yes... that does sound familiar...
Bogus
Every once in a while I wander by to see what's going on; as for the
latest...
Considering that for a while Jim, Tony, and I were the only members of the
Monitor Group (TOPS-10), I can't imagine why people don't think it was
called a monitor.
Also,
Nancy Kilty and Donaleen Kohn wrote and taught the TOPS-20 Monitor
Internals Course; Joluut Vanderhooft was the manager of Large Systems
Course Development
...PMW (on the monitor listings)
...wex
w...@concentric.net
David Harmon
On 15 Apr 1998 17:17:13 GMT, atbo...@thinkage.on.ca (Alan Bowler)
wrote:
> In particular, standard Pascal does
>not allow any form of separate compilation, and the idea of trying to
>maintain a whole compiler as a single huge module module makes me
>shudder.
Shudder away, but the AAEC Pascal compiler mentioned by Richard was
straight Jensen & Wirth Pascal (with concessions to EBCDIC) and compiled
itself from about 6500 lines of source, all one module.
Heinz W. Wiggeshoff
As did XCOM, the compiler for the XPL Compiler Generator system
developed by McKeeman, Horning and Wortman circa 70. And it actually
compiled in less than a 250K partition/region on an IBM System/360.
Heinz W. Wiggeshoff
(lis...@zetnet.co.uk) writes:
> On 1998-04-14 d...@getsetgo.doc.ic.ac.uk said:
> :It was revised as the new dragon book - "Compilers: Principles,
> :Techniques and Tools" by Aho, Sethi and Ullman. Published by
> :Addison-Wesley.
>
> highly recommended. this is the book that enabled us to actually
> understand LR(x) parsing. we still went off and wrote a recursive
> descent parser for the coursework, buuut... ;>
Us? We? It took a nest of 'lisards' to conquer this?
In the summer of 1990, the Carleton University bookstore sold me the
damage-discounted
Compiler Construction Theory and Practice 2nd edition
By William A. Barrett, Rodney M. Bates, David A. Gustafson and
John D. Couch
c/r Science Research Associates, Inc. 1986 (!)
ISBN 0-574-21765-7
(for a mere $60).
The description of LR(n) parser construction for n=0 and 1 was clear
enough to allow me to write a recursive PL/1 program which takes BNF and
writes a parser in the form of explicit state actions (in PL/1, natch),
as opposed to tables.
The book serves as documentation for the Pascal-like QPARSER, which
is somewhat like Yacc (minus the Yucky C output). I have yet to write
QCAD Systems, Inc. for the software and instructor diskette.
There is an extensive bibliography. Strangely, the authors do not
take LR(1) construction to the LALR(1) stage. The rest of the book
covers the usual gamut of compiler course topics: lexical scanning,
some theory, top-down parsing, symbol tables and run-time issues.
jmfb...@ma.ultranet.com
In article <ErJA4...@world.std.com>, m...@world.std.com (Megan) wrote:
>erepa...@cc.curtin.edu.au (Paul Repacholi ( prep )) writes:
>
>>Ah, this is the `monitor' software concept af Hoar et al.
>
>Thanks... yes... that does sound familiar...
I've never heard of s/he/it. Care to elaborate a tad?
/BAH
John Hughes
dgr...@iinet.net.au (David Green) writes:
> On Sun, 12 Apr 98 08:46:29 GMT, Robert Billing
> <uncl...@tnglwood.demon.co.uk> wrote:
>
> Mount Isa Mines in Queensland took delivery of a 1909 in March
> 1966. (Incidently, that 1909 had a 16k 24 bit word memory).
Wow. I NEVER heard of a 1909! I wonder what that was.
> So, is it just a coincidence that George 1 and George 2 were developed
> on Deuce and ported to the English Electric KDF9, and that George 3
> and George 4 were o/s for the 1900 series? Or did the 1950's George
> language metamorphose into the 1960's George o/s?
I guess coincidence - George 1 and George 2 were also O/S's for IC[TL]
1900 machines.
(I never saw much George 2, and no George 1, but I spent 1977-80 at
the University of East Anglia, Norwich, UK using George 3 and 4 on an
ICL 1903T and 1982-83 using them on an ICL 1904S at the University of
Hull, Kingston-upon-Hull, UK).
timo...@cyberramp.net
>This book has emerged from my lecture notes for an introductory
>course in compiler design at ETH Zürich. Several times I have
>been asked to justify this course, since compiler design is
>considered a somewhat esoteric subject, practised only in a few
>highly specialized software houses. Because nowadays everything
>which does not yield immediate profits has to be justified, I
>shall try to explain why I consider this subject as important and
>relevant to computer science students in general.
>--
If you stop teaching how to write compilers, in a few years there will
be nobody left that can dio it.
jmfb...@ma.ultranet.com
In article <3536E112...@concentric.net>,
Bogus <Bo...@concentric.net> wrote:
>Every once in a while I wander by to see what's going on; as for the
>latest...
>
>Considering that for a while Jim, Tony, and I were the only members of the
>Monitor Group (TOPS-10), I can't imagine why people don't think it was
>called a monitor.
You seem to have forgotten Alan and Cliff.
/BAH