Help on Future of VAX / VMS
Terry C Shannon
to pour money into OpenVMS development (e.g TNFS file system, enhanced
clustering, ultimately mixed Unix/VMS clusters) and has no intentions
whatsoever of dumping the OS. Which is just as well--OpenVMS/VAX,
OpenVMS AXP and clustering software are among the top five software
packages that DEC sells.
Unfortunately, DEC has done an utterly squatulent job of marketing and
promoting OpenVMS (gee, why does this not surprise me :-} ).
Furthermore, a plague of OS Political Correctness has rendered it
impossible for the OpenVMS constituency to cast OpenVMS as being "best"
at anything, even when it is. I suspect that DEC now recognizes the
error in its ways and will act accordingly to remove the taint that
currently besmirches the escutcheon of OpenVMS. Fact of the matter is,
if OpenVMS dies, DEC dies. Period. Unix and NT can't pick up the load
in the absence of OpenVMS.
As for the viability of DEC's hardware platforms, the firm clearly wants
to migrate customers from VAX to AXP as soon as practicable.
Nevertheless, new VAX systems are in the pipeline; some may dribble out
as early as next month. And look for a major AXP rollout this Fall. On
deck are an AXP-powered X-windows terminal, an economy-class desktop
system, a 275MHz SuperSable, and the much-vaunted TurboLaser mainframe
killer.
Will AXP be around in 5 or 10 years? We believe it will be. Shipment
volumes are increasing to an acceptable--although not optimal--level and
are expected to continue to do so. DEC did, however, miss a major
opportunity with AXP by refusing to capitalize on the lead they had when
the architecture came out almost two years ago. Again, a good deal of
this is attributable to squatulent marketing. Accordingly, all eyes are
focused on Pentium and PowerPC, neither of which offer AXP levels of
performance.
------------------------------------------------------------------------
Terry C. Shannon 135 Leland Farm Road
Senior Analyst Ashland, MA 01721
Illuminata 508-881-5563
sha...@illuminata.com
Al Brunck
small DEC network. It consists of a VAX 8200 with 3 VT220 terminals, a
VAXstation 2000, and a VAXstation 3100. The system is used to run a custom
software package consisting of ~180K lines of FORTRAN, which includes a
custom database. They are running VMS v5.5-2 and using the FORTRAN
compiler v6.0. They are using v1.9 of Terminal Data Management System
(TDMS) for their user interface (which I understand will not be supported by
DEC in the future). The volume of data which this system processes is fairly
small; additional CPU power is needed only for future growth.
A number of approaches to modernization are being considered. One approach
is to replace the hardware with equivalent hardware being sold today, (ideally)
leaving the software alone. Then, the software could be updated in phases
(e.g., operator interface converted to GUI, use commercial database package).
One thing which concerns me about this approach is the long-term viability of
DEC, the VAX and Alpha architectures, and VMS.
Most of my background is in PCs and embedded systems; thus, I don’t have a
good feel for where the DEC market is headed. Is the VAX architecture dead?
Is the Alpha architecture a viable alternative? Although the Alpha appears to
be technically good, is it going to be around in 10 years? 5 years? What about
an operating system. Is VMS (or Open VMS, as I see it called) a dead end?
Should we go to DEC’s Unix or Windows NT?
Also, any comments on the amount of work involved in porting the FORTRAN
code to another platform and/or OS would be appreciated.
Thanks.
Mark H. Wood
[deletia]
> are expected to continue to do so. DEC did, however, miss a major
> opportunity with AXP by refusing to capitalize on the lead they had when
> the architecture came out almost two years ago. Again, a good deal of
> this is attributable to squatulent marketing. Accordingly, all eyes are
> focused on Pentium and PowerPC, neither of which offer AXP levels of
> performance.
It certainly wouldn't be hard to do better. Every time I turn on the TV I'm
assaulted with Intel ad.s trumpeting the paltry virtues of the Pentium. Where
do I see the Digital logo? The Nightly Business Report -- that's it! There's
the way to build your desktop volume, all right. Just last week I saw, for the
*first time ever*, a local computer store advertising a Digital product.
I certainly *hope* they are going to make improvements. Just leave my sales
rep. in place and get some marketers that know how to take good products to
productive markets, without ruining the product or missing the market.
--
Mark H. Wood, Lead Systems Programmer +1 317 274 0749 [@disclaimer@]
Internet: MW...@INDYVAX.IUPUI.EDU BITNET: MWOOD@INDYVAX
"I live the greatest adventure one could ever wish." - a tosc
Ken Hobday
TDMS. The whole forms area at Digital seems to be headed for
the graveyard, but TDMS in particular is the poor step-child of
the forms group. DECforms is available on OpenVMS AXP and
I believe FMS is as well; however, I don't know of a supported
port of TDMS. In any case, there hasn't been any serious
engineering work done on TDMS for many years.
-Ken
--
Ken Hobday (513) 865-1121 x6764
Mead Data Central 1 Cor. 6:20 Delivery Architecture
P.O. Box 933 "Bought with a price" ke...@meaddata.com
Dayton, Ohio 45401 ...!uunet!meaddata!kenh
Jerry Leslie
: Probably the biggest problem your client faces is what to do about
Here's an article on TDMS not being ported to the ALPHA...
--Gerald (Jerry) R. Leslie
Staff Engineer
Dynamic Matrix Control Corporation (my opinions are my own)
P.O. Box 721648 9896 Bissonnet
Houston, Texas 77272 Houston, Texas, 77036
713/272-5065 713/272-5200 (fax)
gle...@isvsrv.enet.dec.com
jle...@dmccorp.com
==============================================================================
From: vanden...@eps.enet.dec.com (Hein RMS van den Heuvel)
Newsgroups: comp.os.vms
Subject: Re: TDMS for the Alpha???
Date: 7 JUL 94 00:59:07
Organization: Digital Equipment Corporation
Lines: 36
Message-ID: <2vg2ks$8...@jac.zko.dec.com>
References: <joe.hayashibar...@grimm.nb.rockwell.com>
NNTP-Posting-Host: EPSYS
In article <joe.hayashibar...@grimm.nb.rockwell.com>, joe.hay...@nb.rockwell.com (Joe Hayashibara) writes...
>
> Does anybody know when TDMS will be release for OpenVMS AXP???
VAX TDMS is not being ported to Alpha AXP. This forms product was put into
maintenance-only mode over four years ago when its replacement product,
DECforms, was first released. Through a joint project between Digital
and Praxa Limited, a Digital Business Partner in Melbourne, Australia,
two conversion tools are now available to assist customers in migrating
their VAX TDMS applications from VAX to Alpha AXP.
The first is the VAX TDMS to DECforms Converter for OpenVMS VAX which
converts VAX TDMS forms and requests to DECforms forms. Included with this
package is a COBOL Translator which converts the TDMS calls in COBOL
applications to DECforms calls. Once converted to DECforms, the application and
forms can be ported to Alpha AXP. (QB-270AA-xx TDMStoForms CNV)
The second tool is the VAX TDMS Emulator for OpenVMS which supports the
conversion of VAX TDMS forms and requests to a 3GL environment which uses
OpenVMS SMG screen management routines to emulate the VAX TDMS environment.
The resultant application can then be ported to Alpha AXP.
(QB-26BAA- xx TDMS EMUL DV VMS)
To order and/or to receive further information you may want to contact
either your Digital representative or:
Jay Ondracek
Praxa Limited
South Melbourne, Australia
International Phone: 011-61-3-690-3811
Internet Address: ondr...@praxa.com.au
Hope this helps, +--------------------------------------+
| All opinions expressed are mine, and |
Hein van den Heuvel | may not reflect those of my employer |
vanden...@eps.enet.dec.com +--------------------------------------+
Douglas Miller
a...@xetron.com (Al Brunck) writes:
>Most of my background is in PCs and embedded systems; thus, I don’t have a
>good feel for where the DEC market is headed. Is the VAX architecture dead?
No, but is is dying --- DEC are cutting down the VAX range as they
expand the AXP (Alpha based) range.
>Is the Alpha architecture a viable alternative?
Of course, that is what it was designed for.
>Although the Alpha appears to
>be technically good, is it going to be around in 10 years? 5 years?
I'd put money on it. Look at the facts: DEC has a long track-record
of engineering excellence; when the first Alpha came out it was the
world's fastest microprocessor; DEC claim that the architectire will
last at least 25 years and be scaled to 1000 times initial
performance.
>What about an operating system. Is VMS (or Open VMS, as I see it
>called) a dead end?
No. VMS is a mature operating system that is exceptionally
functional, robust, and standards-compliant. The very opposite of a
dead-end in fact.
>Should we go to DEC's Unix or Windows NT?
You don't have to go anywhere to get "Unix" --- VMS is "Unix" for any
practical purpose. Of course some people wil try and tell you that
something isn't *really* Unix unless it can trace its source code
lineage back to thde Holy Fount at AT&T. You can ignore these bigots
of course, because the only thing that counts is what an operating
system *does*, not where it come from. If you want "Unix"
functionality, then specify the formal standards: POSIX. VMS is more
POSIX compliant than many of its competitors which claim to be "Unix".
Howard C. Smith
Miller) writes:
> You don't have to go anywhere to get "Unix" --- VMS is "Unix" for any
????? VMS is UNIX ?????
The management at DEC must have shipped their whole sales force to
this poor naive user to convince him of this one! Unfortunately for VMS
users, there is a lot more to UNIX than just POSIX compliance.
--
Howard C. Smith
sm...@nextone.niehs.nih.gov
Stephen J. Anderson
In a previous article, dou...@brt.deakin.edu.au (Douglas Miller) says:
>...
>You don't have to go anywhere to get "Unix" --- VMS is "Unix" for any
>practical purpose. Of course some people wil try and tell you that
>something isn't *really* Unix unless it can trace its source code
>...
I was willing to let the previous drivel about VMS being "exceptionally
functional, robust, and standards-compliant", go untorched, however, saying
VMS is UNIX for _any_ purposes is complete rubbish. VMS is awkard,
inflexible, expensive, lacks reliable development tools, and has a list of
DEC-no-longer-supported software that rivals the supported software list.
No, I won't get into DECnet Phase V.
If DEC is your supplier of choice, then alpha is the direction to go, and
from most accounts I have read, OSF/1 performs reasonably well.
obe...@icaen.llnl.gov
av...@FreeNet.Carleton.CA (Stephen J. Anderson) writes:
>I was willing to let the previous drivel about VMS being "exceptionally
>functional, robust, and standards-compliant", go untorched, however, saying
>VMS is UNIX for _any_ purposes is complete rubbish. VMS is awkard,
>inflexible, expensive, lacks reliable development tools, and has a list of
>DEC-no-longer-supported software that rivals the supported software list.
>No, I won't get into DECnet Phase V.
>
>If DEC is your supplier of choice, then alpha is the direction to go, and
>from most accounts I have read, OSF/1 performs reasonably well.
I am tempted to flame, but those who have no clue are not going to discover
reality on USENET.
Unix is a trademark of X/Open and it is clearly defined by them. In these legal
terms, VMS IS UNIX. This is, obviously, pure legal silliness. From any
practical perspective VMS is not Unix or even close to it. But as long as it
meets the letter of the law (X/Open's requirements for Unix branding) it is
legally Unix and VMS can be bid on government contracts calling for Unix.
This is not a trivial issue!
As to the list of silly and ignorant claims against VMS, I doubt that you will
get a clue. But the lawyers probably won't, either.
R. Kevin Oberman
Energy Sciences Network (ESnet)
National Energy Research Supercomputer Center (NERSC)
Lawrence Livermore National Laboratory (LLNL)
Internet: kobe...@llnl.gov +1 510-422-6955
Roger Barnett
>
> In article <31ljo9$k...@sol.ccs.deakin.edu.au> dou...@brt.deakin.edu.au
(Douglas
> Miller) writes:
>
> The management at DEC must have shipped their whole sales force to
> this poor naive user to convince him of this one! Unfortunately for VMS
> users, there is a lot more to UNIX than just POSIX compliance.
Which explains why Unix <> Open Systems ?
--
Roger Barnett
"Where Time tonight may find the time to kill"
Clive James
Peter da Silva
Douglas Miller <dou...@brt.deakin.edu.au> wrote:
> You don't have to go anywhere to get "Unix" --- VMS is "Unix" for any
> practical purpose.
No, VMS is "POSIX". POSIX is a useful spec for writing portable applications,
if your application doesn't have to know much about its environment. It doesn't
address applications, tools, system administrations issues, networking, or any
number of similar things that you have to know about.
For example, look at "C-NEWS". That's a pretty portable application, but it
knows all *sorts* of things about the system.
> Of course some people wil try and tell you that
> something isn't *really* Unix unless it can trace its source code
> lineage back to thde Holy Fount at AT&T.
Linux is "unix". MINIX is "unix". Neither of these have line 1 of AT&T
code in them.
> You can ignore these bigots
> of course, because the only thing that counts is what an operating
> system *does*, not where it come from.
Exactly. Open systems are a matter of interfaces and protocols. VMS provides
some of the interfaces that define UNIX. It doesn't provide all of them.
What happens on your VMS machine when I type "crontab -e".
> If you want "Unix"
> functionality, then specify the formal standards: POSIX.
Formal standards are nice for justifying a purchase request to the bean
counters. Practical considerations are more important, however, when you
want to get work done.
And here I'm going to shock a whole shitload of people. I'm not going to
say "don't buy VMS". I'm going to say: look at your workload. Look at your
requirements. Buy a system that satisfies those requirements. The only reason
to buy a "unix" system is if you need some facility you can't get outside
of a "unix" system. If you can get all the facilities you need on a VMS
system, go for it. But one facility you won't get on VMS is the ability to
compile and run most "unix" software packages. That's because it's not "unix".
Just as you won't be able to run most VMS programs under "unix". That's
because it's not VMS.
Now at this point someone is going to come in with a list of a dozen "unix"
programs that have been ported to VMS, or a dozen "unix" programs that are
actually POSIX programs so run under VMS.
I can come back with a dozen more that aren't, or for which the VMS version
is behind the "unix" one. The point is... look at what *you* want to run, and
get a system that does the job *you* want.
A more important question is... what *applications* do I buy. Because it's
the applications that are going to determine whether you're locked into an
operating system or not.
I would personally recommend against buying applications that depend on the
VMS environment in any fundamental way, if there's an alternative. But that's
just me... I'd rather be locked into "unix" than VMS.
--
Peter da Silva `-_-'
Network Management Technology Incorporated 'U`
1601 Industrial Blvd. Sugar Land, TX 77478 USA
+1 713 274 5180 "Hast Du heute schon Deinen Wolf umarmt?"
jg_sun
>> Also, any comments on the amount of work involved in porting the FORTRAN
>> code to another platform and/or OS would be appreciated.
It really depends on how dependent you program on the RMS and all other
VMS system calls. We have a data analysis package which made VMS system
calls only to trap errors and calculate cpu time, and the package
is ported to Unix without much error. However, we also have a package
which use the RMS and FMS extensively, and nobody dare to think about
port the damn thing anywhere. So if you stuck with VMS, you stuck.
Choosing between VMS and Unix is more an emotional issue. If you
depend heavily on VMS system calls, it is going to be extremly painful
to port to ANYWHERE, including Unix, NT etc., much much more difficult
than porting between different Uni*s, so for practical reason
it may be easier follow OpenVMS, if you have enough money. There
are plenty of people stuck like you, so VMS may live for another
10 years.
j.g.
obe...@icaen.llnl.gov
obe...@icaen.llnl.gov writes:
>Unix is a trademark of X/Open and it is clearly defined by them. In these legal
>terms, VMS IS UNIX. This is, obviously, pure legal silliness. From any
>practical perspective VMS is not Unix or even close to it. But as long as it
>meets the letter of the law (X/Open's requirements for Unix branding) it is
>legally Unix and VMS can be bid on government contracts calling for Unix.
>
>This is not a trivial issue!
Time for a correction. I was getting Unix branding and POSIX confused. VMS is
officially POSIX conformant and this is what is required to bid on "Unix"
proposals. While I'm not sure of the status of X/Open's Unix branding, it is
not inconcevable that VMS will bet it and be legally, if net any other way,
Unix.
Mea culpa.
I'm surprised that I have only received one message correcting my error.
Jerry Leslie
: j.g.
There are packages that implement RMS & other VMS hooks; e.g., RMS
for OSF/1 from Accelr8 Technology. The following is what they've announced
for the ALPHA, as of January, 1994.
--Gerald (Jerry) R. Leslie
Staff Engineer
Dynamic Matrix Control Corporation (my opinions are my own)
P.O. Box 721648 9896 Bissonnet
Houston, Texas 77272 Houston, Texas, 77036
713/272-5065 713/272-5200 (fax)
gle...@isvsrv.enet.dec.com
jle...@dmccorp.com
==============================================================================
ALPHA AXP APPLICATIONS DETAIL REPORT Page: 6
5-JAN-1994
SOFTWARE VENDOR OPR ANNOUNCED
APPLICATIONS SYS SHIP DATE
-------------------------------------------------------------------------
Accelr8 Technology Corporation
303 E. 17th Ave.
Suite 108
Denver, CO United States 802031253
Contact: Timothy M Fitzpatrick
Phone: (303) 863-8088
D-ISAM V3.20 NT JAN 1995
D-ISAM V3.20 OSF SHIPPING NOW
D-ISAM V3.20 VMS SHIPPING NOW
Dev Kit V1.0 NT QTR2 CY1994
Dev Kit V1.0 OSF QTR2 CY1994
Open ACCLIM8 V1.3 NT JAN 1995
Open ACCLIM8 V1.3 OSF SHIPPING NOW
Open BASIC V1.0 NT JAN 1995
Open BASIC V1.0 OSF SHIPPING NOW
Open BASIC V1.0 VMS QTR4 CY1993
Open DCL V1.10 NT JAN 1995
Open DCL V1.10 OSF SHIPPING NOW
Open INTEGR8 V1.0 NT JAN 1995
Open INTEGR8 V1.0 OSF SHIPPING NOW
Open INTEGR8 V1.0 VMS SHIPPING NOW
Open LIBR8 V2.01 NT JAN 1995
Open LIBR8 V2.01 OSF SHIPPING NOW
Open MIGR8 NT QTR4 CY1994
Open MIGR8 OSF SHIPPING NOW
Open MIGR8 VMS SHIPPING NOW
Open RMS V2.01 NT JAN 1995
Open RMS V2.01 OSF SHIPPING NOW
Open SMG V1.0 NT JAN 1995
Open SMG V1.0 OSF SHIPPING NOW
Open TPU V1.0 NT JAN 1995
Open TPU V1.0 OSF SHIPPING NOW
Open TRANSL8 V2.50 NT JAN 1995
Open TRANSL8 V2.50 OSF SHIPPING NOW
Open TRANSL8 V2.50 VMS SHIPPING NOW
Peter da Silva
Roger Barnett <Ro...@natron.demon.co.uk> wrote:
> > The management at DEC must have shipped their whole sales force to
> > this poor naive user to convince him of this one! Unfortunately for VMS
> > users, there is a lot more to UNIX than just POSIX compliance.
> Which explains why Unix <> Open Systems ?
No, it's why POSIX <> Complete API.
James R. Bull Jr
>which isn't all that different from on a Ultrix machine: :)
>
># crontab -e
>crontab: Command not found.
>
which is why ultrix <> unix or at least ultrix < unix
James Bull
VMS and Ultrix SysAdmin
State of Oregon
Water Resources Department
Salem, Oregon
Chris Faylor
jg_sun <js...@uvahev.phys.Virginia.EDU> wrote:
>In article <Ctnzn...@xetron.com> a...@xetron.com (Al Brunck) writes:
>
>>> code to another platform and/or OS would be appreciated.
>
>VMS system calls. We have a data analysis package which made VMS system
>which use the RMS and FMS extensively, and nobody dare to think about
>port the damn thing anywhere. So if you stuck with VMS, you stuck.
Actually there are companies out there who provide RMS packages for UNIX.
This is one:
Sector7
5802 Lakeview Circle
Austin, TX 78731
Tel: 512-451-3961
FAX: 512-451-8643
--
Chris Faylor Boston Business Computing, Ltd.
c...@ednor.bbc.com "I feel more like I do now than I did when I first got here"
Peter da Silva
John Hascall <jo...@iastate.edu> wrote:
> Peter da Silva <pe...@nmti.com> wrote:
> }What happens on your VMS machine when I type "crontab -e".
> $ crontab -e
> %DCL-W-IVVERB, unrecognized command verb - check validity and spelling
> \CRONTAB\
> which isn't all that different from on a Ultrix machine: :)
> # crontab -e
> crontab: Command not found.
Ha ha. Funny man.
Next question, which one is it easier to install one of the free System V
compatible crontab programs on?
g...@brt.deakin.edu.au
pe...@nmti.com (Peter da Silva) writes:
>In article <31ljo9$k...@sol.ccs.deakin.edu.au>,
>Douglas Miller <dou...@brt.deakin.edu.au> wrote:
>> You don't have to go anywhere to get "Unix" --- VMS is "Unix" for any
>> practical purpose.
>
>No, VMS is "POSIX". POSIX is a useful spec for writing portable applications,
>if your application doesn't have to know much about its environment. It doesn't
>address applications, tools, system administrations issues, networking, or any
>number of similar things that you have to know about.
>
>For example, look at "C-NEWS". That's a pretty portable application, but it
>knows all *sorts* of things about the system.
>
>> Of course some people wil try and tell you that
>> something isn't *really* Unix unless it can trace its source code
>> lineage back to thde Holy Fount at AT&T.
>
>Linux is "unix". MINIX is "unix". Neither of these have line 1 of AT&T
>code in them.
>
>> You can ignore these bigots
>> of course, because the only thing that counts is what an operating
>> system *does*, not where it come from.
>
>Exactly. Open systems are a matter of interfaces and protocols. VMS provides
>some of the interfaces that define UNIX. It doesn't provide all of them.
>
$ posix
psx> crontab -e
it starts up vi and edits my crontab file. what did you expect it to do?
:-)
Better check you facts first, Peter.
Greg O'Sullivan
(g...@brt.deakin.edu.au)
...stuff deleted
Peter da Silva
Roger Barnett <Ro...@natron.demon.co.uk> wrote:
> For example, what is the relative importance to the user (as opposed
> to the marketing division) of: Posix, XPG4, Spec1170 ?
Posix is almost meaningless. Posix is something yu use when writing an
application for portability. It gives you an idea how much work porting
your application is going to be, based on how many times yu have to violate
Posix (yes, you *will* have to).
XPG4 is better.
Spec1170 right now means Unixware. Whether it is a useful spec is going to
have to wait until the second Spec1170 system is out.
They're all relatively useless. Better questions, depending on the applications
you're interested in are:
Does it run SCO binaries? Unixware binaries? Solaris binaries?
Does it satisfy the SVID?
Does it support the BSD API (native sockets, for example).
Paul Ockenden
> extensively, and nobody dare to think about port the damn
> thing anywhere. So if you stuck with VMS, you stuck.
We specialise in porting such applications to Unix or NT.
Things like this CAN be done. We even have a portable SMG$
emulation library which we had to write to get an app running
under NT !!!!
So don't give up hope - no-one is *stuck* with VMS....
P.
--------------------------------------------------------------------
Paul Ockenden, Managing Director, Industrial Might & Logic Limited
58 Cobden Road, BRIGHTON, East Sussex, BN2 2TJ, U.K.
Phone: +44 273 621393 Fax: +44 273 621390 Mobile: +44 850 626206
--------------------------------------------------------------------
Peter da Silva
> >What happens on your VMS machine when I type "crontab -e".
> $ posix
> psx> crontab -e
> it starts up vi and edits my crontab file. what did you expect it to do?
Give me the huge spammy error message John Hascall came up with.
I'm somewhat impressed.
OK, let's try something harder. My first example:
tar xvf cnews.tar
cd conf
build
...
I've ported C news to Xenix-286. It wasn't even hard... apart from wrong
guesses as to what was in the runtime libraries I don't think I had to touch
it. I would expect C News to install pretty much mindlessly on any UNIX
system. If it doesn't, then you've got a pretty broken UNIX... as proven by
my installing it *on* as broken a UNIX as I know of.
If OpenVMS is less UNIXy then Xenix-286, it's probably not UNIX at all.
--
Luke Brennan
> Peter da Silva <pe...@nmti.com> wrote:
>
> %DCL-W-IVVERB, unrecognized command verb - check validity and spelling
> \CRONTAB\
>
> which isn't all that different from on a Ultrix machine: :)
>
> # crontab -e
> crontab: Command not found.
How about my HPUX 9.05 box? :-} (take it - please!)
$ crontab -e
crontab: you are not authorized to use crontab. Sorry.
$ su
# crontab -e
crontab: can't open your crontab file.
I look in the man page - no mention of '-e' as an option.
So what is Peter's point??? (he should get VMS, then? :-))
Luke.
Dan Pop
>John Hascall <jo...@iastate.edu> writes:
>
>>which isn't all that different from on a Ultrix machine: :)
>>
>># crontab -e
>>crontab: Command not found.
>>
>
>which is why ultrix <> unix or at least ultrix < unix
By your argument, any BSD-based Unix is not Unix because it doesn't
have some SystemV commands.
In BSD, there is only one crontab file, controlled by root, and
the entries have an extra field which specifies under which user
the process will run.
Dan
--
Dan Pop
CERN, CN Division
Email: dan...@cernapo.cern.ch
Mail: CERN - PPE, Bat. 31 R-004, CH-1211 Geneve 23, Switzerland
Douglas Miller
>> You don't have to go anywhere to get "Unix" --- VMS is "Unix" for any
>
> ????? VMS is UNIX ?????
Yes, you read right the first time. This fact seems to really upset
some people...
>The management at DEC must have shipped their whole sales force to
>this poor naive user
"...sales force", "poor", "naive", "user". How many gratutious
insults can you fit into one sentence? :-)
>to convince him of this one! Unfortunately for VMS
>users, there is a lot more to UNIX than just POSIX compliance.
Posix *is* Unix by definition. The whole purpose of the Posix effort
is to distill a consensus on what "unix" really means into a set of
formal standards that define *function* not implementation. These
standards can then be implemented on a wide variety of operating
systems, including those such as VMS whose internals are radically
different to what has traditioanlly been thought of as "Unix".
Douglas Miller
>
>>...
>>You don't have to go anywhere to get "Unix" --- VMS is "Unix" for any
>>practical purpose. Of course some people wil try and tell you that
>>something isn't *really* Unix unless it can trace its source code
>>...
>I was willing to let the previous drivel about VMS being "exceptionally
>functional, robust, and standards-compliant", go untorched,
But then you thought of a way of getting in a sneaky "en passant"
attack.
>however, saying VMS is UNIX for _any_ purposes is complete rubbish.
Why? If the following is intended to support your assertion then it
is a non-secitur.
> VMS is awkard, inflexible, expensive, lacks reliable development
> tools,
Your opinions. All of these opinions are also held by other people
about other unix operating systems (in various combinations). You
have no point.
>and has a list of
>DEC-no-longer-supported software that rivals the supported software list.
Yes DEC have no-longer supported software. Big Deal. Like any
company, they don't sell and support all products forever. Again you
have no point.
>No, I won't get into DECnet Phase V.
Just as well probably; pointless attacks are best kept short.
Douglas Miller
pe...@nmti.com (Peter da Silva) writes:
>In article <31ljo9$k...@sol.ccs.deakin.edu.au>,
>Douglas Miller <dou...@brt.deakin.edu.au> wrote:
>> practical purpose.
>
And posix defines "Unix".
>POSIX is a useful spec for writing portable applications,
>if your application doesn't have to know much about its environment.
Kind of follows from the definition of "portable".
>It doesn't
>address applications, tools, system administrations issues, networking, or any
>number of similar things that you have to know about.
You seem to be regarding Posix as just Posix.1 (system calls) and
Posix.2 (shell and utilities). Although these are the most basic,
longest-standing, and most widely implemented, there are a raft of
others in various states of completion that address some of the other
things you mention (among others). But what are "applications and
tools"? Things you feel shgould have been in Posix.2?
>For example, look at "C-NEWS". That's a pretty portable application, but it
>knows all *sorts* of things about the system.
>
>> Of course some people wil try and tell you that
>> something isn't *really* Unix unless it can trace its source code
>> lineage back to thde Holy Fount at AT&T.
>
>Linux is "unix". MINIX is "unix". Neither of these have line 1 of AT&T
>code in them.
Yes I know. But are they Unix? My point was that people who say
"yes" will say so because they feel Linux and Minix share a common
ancestor in AT&T Unix (this is what I mean by "source code lineage"
--- I'm not implying that there has to be any common source code).
>> You can ignore these bigots
>> of course, because the only thing that counts is what an operating
>> system *does*, not where it come from.
>
>Exactly. Open systems are a matter of interfaces and protocols. VMS provides
>some of the interfaces that define UNIX. It doesn't provide all of them.
But the purpose of Posix is to define the "Unix" interfaces! The
Posix standards in their current state of development are the only
definition of Unix that we have (until/unless spec 1170 gets going).
>What happens on your VMS machine when I type "crontab -e".
In compliance with Posix.2, it lets you edit your crontab entry! Hmm,
maybe Posix is not quite as useless as you thought?
>Formal standards are nice for justifying a purchase request to the bean
>counters.
But you are doing a diservice to yourself and your bean-counters if
the formal standards specified in your purchase request don't service
a real pratical need.
>Practical considerations are more important, however, when you
>want to get work done.
Your premise is that formal standards are by definition impractical
(obviously nonsense), and hence Posix is impractical. To justify this
you'll have to do better that some arm-waving about missing features
--- how about some specifics (and remember you've already been wrong
about crontab :-).
>Now at this point someone is going to come in with a list of a dozen "unix"
>programs that have been ported to VMS, or a dozen "unix" programs that are
>actually POSIX programs so run under VMS.
>
>I can come back with a dozen more that aren't,
Let me make some educated guesses/predictions
o the ones that don't also don't work on plenty of other "unix"
systems.
o the ones that do, work on a greater range of systems than the ones
that don't.
I.e., Posix is a Good Thing for portable applications.
Douglas Miller
obe...@icaen.llnl.gov writes:
>Time for a correction. I was getting Unix branding and POSIX confused. VMS is
>officially POSIX conformant and this is what is required to bid on "Unix"
>proposals.
Well, if the proposal defines "Unix" that way. What I suspect is that
a lot of the time "unix" is not defined clearly or at all (kind of
goes with the kind of sloppy thinking that produces proposals with the
vague term "Unix" as a key part), and hence have to accept Posix
compliant offerings as perfectly adequate proma facie proposals. I've
heard of DEC actually winning at least one tender in this way with
VMS, much to the suprise of the drafters of the tender document.
>While I'm not sure of the status of X/Open's Unix branding,
It is called XPG (in simple terms it is Posix plus a compliant C
compiler).
>it is not inconceivable that VMS will bet it
VMS has been XPG branded for years.
Douglas Miller
pe...@nmti.com (Peter da Silva) writes:
>Posix is almost meaningless. Posix is something yu use when writing an
>application for portability. It gives you an idea how much work porting
>your application is going to be, based on how many times yu have to violate
>Posix (yes, you *will* have to).
Only for a target system that is not fully compliant; then it is that
system that is violating Posix, not you.
>XPG4 is better.
For C programmers.
>Spec1170 right now means Unixware. Whether it is a useful spec is going to
>have to wait until the second Spec1170 system is out.
>
>They're all relatively useless. Better questions, depending on the applications
>you're interested in are:
>
> Does it run SCO binaries? Unixware binaries? Solaris binaries?
> Does it satisfy the SVID?
> Does it support the BSD API (native sockets, for example).
Now we are really in looking-glass land. Vendor-independent
interfaces are "useless", and proprietary interfaces are worthwhile?!
A plank, a stick and a sheet
> So what is Peter's point??? (he should get VMS, then? :-))
Probably to show us all what a yfeefy he is.
Tom O'Toole - ecf_...@jhuvms.hcf.jhu.edu - JHUVMS system programmer
Homewood Computing Facilities - Johns Hopkins University, Balto. Md. 21218
Arrr, there's a worthy vessel! Surely this is the yarest craft I've ever set
eyes on... Arrr, I don't know what I'm talking about... - the 'captain'
Andreas Fassl
>Peter da Silva <pe...@nmti.com> wrote:
>}What happens on your VMS machine when I type "crontab -e".
When YOU type "crontab -e" on MY openVMS-system will happen nothing at all :-)
BUT
>$ crontab -e
>%DCL-W-IVVERB, unrecognized command verb - check validity and spelling
> \CRONTAB\
can't work, you have to be in the posix mode. The posix license is part
of openVMS, you only have to install it.
With kind regards
Andreas
--
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
+ proGIS Softwareentwicklung, Simulationssysteme, Beratung +
+ Germany - 52064 Aachen, Jakobstrasse 181 +
+ E-Mail: and...@didymus.rmi.de VOICE: (49) 241 403 446 +
Peter da Silva
Douglas Miller <dou...@brt.deakin.edu.au> wrote:
> In Article <id.K8T...@nmti.com>
> pe...@nmti.com (Peter da Silva) writes:
> >In article <31ljo9$k...@sol.ccs.deakin.edu.au>,
> >Douglas Miller <dou...@brt.deakin.edu.au> wrote:
> >> You don't have to go anywhere to get "Unix" --- VMS is "Unix" for any
> >> practical purpose.
> >No, VMS is "POSIX".
> And posix defines "Unix".
See previous message. POSIX doesn't define UNIX. In fact it's the least
restrictive subset of UNIX you can get (see also, XPGn, Spec 1170, ...).
In fact X/Open, which now owns the name UNIX, doesn't even think their
own *more restrictive* spec is yet sufficient.
> >POSIX is a useful spec for writing portable applications,
> >if your application doesn't have to know much about its environment.
> Kind of follows from the definition of "portable".
No. A portable application is one that know no more about its environment
than it needs to. Not one that doesn't know anything about its environment.
That's like saying a portable computer is one that weighs nothing and
occupies no space, since otherwise you can't take it with you when you're
surfing.
> You seem to be regarding Posix as just Posix.1 (system calls) and
> Posix.2 (shell and utilities).
Well, it's awfully difficult to be compliant with a standard that isn't
yet finalised.
> >Linux is "unix". MINIX is "unix". Neither of these have line 1 of AT&T
> >code in them.
> Yes I know. But are they Unix?
Yes.
> My point was that people who say
> "yes" will say so because they feel Linux and Minix share a common
> ancestor in AT&T Unix
I should hope they don't. That's a good way to get sued.
But, like POSIX, they are modelled on UNIX. They just happen to implement a
lot of details that fell through the cracks in POSIX.
> >> You can ignore these bigots
> >> of course, because the only thing that counts is what an operating
> >> system *does*, not where it come from.
> >
> >Exactly. Open systems are a matter of interfaces and protocols. VMS provides
> >some of the interfaces that define UNIX. It doesn't provide all of them.
> But the purpose of Posix is to define the "Unix" interfaces!
Just because a standard is intended to perform a particular function does
not mean that it will acheive that goal. After all, SVID was also supposed
to define the "Unix" interfaces, as was OSF/1. In a way the Software Tools
project did the same thing, at yet a lower level. All of these have failed
to provide a sufficiently complete API that is widely enough implemented to
do the job. And even the most complete one, SVID, still allowed a vendor to
ship us broken software and refuse to fix it on the grounds that it *did*
technically satisfy the SVID.
Now don't get me wrong here (it's too late! He did) I'd *love* to see VMS
provide a complete UNIX operating environment. POSIX, however, is not
complete enough to satisfy me. Therefore a POSIX VMS does not by itself
qualify as UNIX in my book.
> >Formal standards are nice for justifying a purchase request to the bean
> >counters.
> But you are doing a diservice to yourself and your bean-counters if
> the formal standards specified in your purchase request don't service
> a real pratical need.
Why, of course! That's why I would never specify POSIX.
> >Practical considerations are more important, however, when you
> >want to get work done.
> Your premise is that formal standards are by definition impractical
> (obviously nonsense), and hence Posix is impractical.
No, my premise is that POSIX is incomplete. In response to your implication
that formal standards are by definition complete what can I say but that
it isn't so.
> --- how about some specifics (and remember you've already been wrong
> about crontab :-).
OK. Port CNEWS to OpenVMS. Provide justification for any modification to
any file outside of the "conf" directory, as to why that modification would
be necessary on a UNIX system. C News is one of the more portable programs
I have worked on. Henry Spencer did an excellent job.
> I.e., Posix is a Good Thing for portable applications.
I entirely agree! In spades! In fact I just said that in my last message!
It is, however, incomplete (and in some areas, for example job control,
probably broken. I won't ding it for job control, though... I don't
believe that belongs in the spec in the first place).
Peter da Silva
Douglas Miller <dou...@brt.deakin.edu.au> wrote:
> In Article <id._9V...@nmti.com>
> pe...@nmti.com (Peter da Silva) writes:
> >Posix is almost meaningless. Posix is something yu use when writing an
> >application for portability. It gives you an idea how much work porting
> >your application is going to be, based on how many times yu have to violate
> >Posix (yes, you *will* have to).
> Only for a target system that is not fully compliant; then it is that
> system that is violating Posix, not you.
No, for a capability that POSIX doesn't provide.
For example, what is the POSIX method of assigning a pty? What is the POSIX
method of finding an unused serial port? What is the POSIX method of disabling
dialin access on a serial port? What is the POSIX method of attaching a
serial port so other applications can't use it? What is the POSIX method
of establishing a TCP connection to a well-known port?
All of these are required for a terminal emulator that can be used for either
local or remote terminal sessions via serial ports or the network. That's not
a particularly oddball application, really... and I seriously doubt it can
be written to straight POSIX, or that you can implement it on OpenVMS without
making calls to VMS.
> > Does it run SCO binaries? Unixware binaries? Solaris binaries?
> > Does it satisfy the SVID?
> > Does it support the BSD API (native sockets, for example).
> Now we are really in looking-glass land. Vendor-independent
> interfaces are "useless", and proprietary interfaces are worthwhile?!
The BSD API is a proprietary interface? Who owns it?
But more to the point: interfaces that you can find substantial software
packages for are more worthwhile than ones you can't. Why else is everyone
so interested in (shudder) WABI?
Peter da Silva
Um, no. By definition UNIX is "Spec 1170". Posix is a series of standards that
each define the API for a subset of UNIX. Posix is a good thing, but it doesn't
really go far enough.
Roger Barnett
> In article <325f4m$a...@sol.ccs.deakin.edu.au>,
> Douglas Miller <dou...@brt.deakin.edu.au> wrote:
> > Posix *is* Unix by definition.
>
> Um, no. By definition UNIX is "Spec 1170". Posix is a series of standards
> that each define the API for a subset of UNIX. Posix is a good thing, but
> it doesn't really go far enough.
Well Digital state that VMS complies approx 85% with the (draft?) Spec1170
document at the moment. How does this compare with the proprietary UNIXes ?
Dan Pop
>BUT
>
>>$ crontab -e
>>%DCL-W-IVVERB, unrecognized command verb - check validity and spelling
>> \CRONTAB\
>
>can't work, you have to be in the posix mode. The posix license is part
>of openVMS, you only have to install it.
Apparently, installing it is not enough in order to be able to use it:
AxCrnC$ posix
psx> uname -a
POSIX_for_OpenVMS_AXP AXCRNC V1.0(V1.0) V1.5-1H1 DEC_3000_Model_500 Alpha
psx> pwd
/43dkc300/danpop
psx> crontab -e
crontab: not appropriate privileges
psx> echo blah >junk
cannot open "junk" for output: Read-only file system
psx> ls -ld .
drwxr-xr-x 2 danpop l3_1 2048 Jul 20 1992 .
Needless to say, from DCL I can write files on my home directory.
And I'm not the sysadmin of the cluster, so I can't fix it, unless
it can be done with luser privileges.
Douglas Miller
>Douglas Miller <dou...@brt.deakin.edu.au> wrote:
>> In Article <id._9V...@nmti.com>
>> pe...@nmti.com (Peter da Silva) writes:
>> >Posix is almost meaningless. Posix is something yu use when writing an
>> >application for portability. It gives you an idea how much work porting
>> >your application is going to be, based on how many times yu have to violate
>> >Posix (yes, you *will* have to).
>
>> Only for a target system that is not fully compliant; then it is that
>> system that is violating Posix, not you.
>
>No, for a capability that POSIX doesn't provide.
OK, terminology problem --- I don't think "violating" is correct. If
you had instead said that "Posix doesn't provide all the interfaces
you will need for every application", then I would have agreed.
>For example, what is the POSIX method of assigning a pty? What is the POSIX
>method of finding an unused serial port? What is the POSIX method of disabling
>dialin access on a serial port? What is the POSIX method of attaching a
>serial port so other applications can't use it? What is the POSIX method
>of establishing a TCP connection to a well-known port?
Well in VMS the pty's are under /dev as you would expect, and there is
socket support for TCP (and IP and UDP), but beyond that I don't know
enough Posix to comment in detail. You've obviously studied Posix in
some depth --- please tell us the answers.
Dan Pop
>In Article <id._9V...@nmti.com>
>pe...@nmti.com (Peter da Silva) writes:
>
>>Posix is almost meaningless. Posix is something yu use when writing an
>>application for portability. It gives you an idea how much work porting
>>your application is going to be, based on how many times yu have to violate
>>Posix (yes, you *will* have to).
>
>Only for a target system that is not fully compliant; then it is that
>system that is violating Posix, not you.
Then, could you show us a little program, which uses TCP/IP networking
without violating POSIX?
>>
>>They're all relatively useless. Better questions, depending on the applications
>>you're interested in are:
>>
>> Does it run SCO binaries? Unixware binaries? Solaris binaries?
>> Does it satisfy the SVID?
>> Does it support the BSD API (native sockets, for example).
>
>Now we are really in looking-glass land. Vendor-independent
>interfaces are "useless", and proprietary interfaces are worthwhile?!
If you can't program your _non-trivial_ application using
vendor-independent interfaces, but you can with "proprietary"
interfaces, then the answer to your question is: YES.
Steve J Hanselman
C'mon guys, what a stupid discussion, why not argue about the colour of the
manuals or something, we're all propogating the myth.
Customers should really look at it this way:-
1. Will my users ever see the operating system
(eg to a user wp/motif, wp/windows et all are all fairly similar,
but are they open?)
2. Will I develop software, if not then can I buy all the software I
need on the hardware/operating system I have chosen?
And can I get contract staff/staaff with relevant experience
at a good price?
3. And for us software developers, can I write my system in a modular
fashion such that I can move it easily.
And is the software development environment good enough to support
productive programming. Eg we often develop unix apps under VMS
as the environment is better.
Just because it runs on Unix doesn't make it portable, when we moved some of
our software from an Ultrix box to a SunSparc, we were horrified to find
only 12 function keys above the keyboard!!
--
Steve J Hanselman
Ste...@brendata.demon.co.uk | I went to a fight
+44 (0268) 490280 | and an Ice Hockey match broke out
Laindon, Essex. UK
Douglas Miller
pe...@nmti.com (Peter da Silva) writes:
>> >POSIX is a useful spec for writing portable applications,
>> >if your application doesn't have to know much about its environment.
>
>> Kind of follows from the definition of "portable".
>
>No.
Yes. I'm saying you have made a tautology.
>A portable application is one that know no more about its environment
>than it needs to.
Yes, that is what you said, and I agree with.
>Not one that doesn't know anything about its environment.
>That's like saying a portable computer is one that weighs nothing and
>occupies no space, since otherwise you can't take it with you when you're
>surfing.
Amusing, but pointless as neither of this have claimed this.
>> >Linux is "unix". MINIX is "unix". Neither of these have line 1 of AT&T
>> >code in them.
>
>> Yes I know. But are they Unix?
>
>Yes.
Why do you think this (I'm curious)?
>> My point was that people who say
>> "yes" will say so because they feel Linux and Minix share a common
>> ancestor in AT&T Unix
>
>I should hope they don't. That's a good way to get sued.
By the Thought Police? I don't think even in the US anyone has worked
out how to sue someone for how they feel.
>But, like POSIX, they are modelled on UNIX. They just happen to implement a
>lot of details that fell through the cracks in POSIX.
Hold on you said they *were* Unix, now you say they are *modelled* on
Unix, i.e., "Unix" is something different.
>Just because a standard is intended to perform a particular function does
>not mean that it will acheive that goal. After all, SVID was also supposed
>to define the "Unix" interfaces, as was OSF/1. In a way the Software Tools
>project did the same thing, at yet a lower level. All of these have failed
>to provide a sufficiently complete API that is widely enough implemented to
>do the job. And even the most complete one, SVID, still allowed a vendor to
>ship us broken software and refuse to fix it on the grounds that it *did*
>technically satisfy the SVID.
I reckon Posix currently beats all of these as a standard because:
o is more general
o is more vendor neutral
o is more widely adopted
Now you reckon that Posix is inadequate. I am not going to disagree
(after all it is still under development). I sure no-one is claiming
that it a panacea for the portability problem. What I am saying that
Posix is the best definition of "Unix" that we have got.
>Now don't get me wrong here (it's too late! He did) I'd *love* to see VMS
>provide a complete UNIX operating environment. POSIX, however, is not
>complete enough to satisfy me. Therefore a POSIX VMS does not by itself
>qualify as UNIX in my book.
The implication being that you have a definition of "Unix" that is
better than just "Posix". What is your definition? I am really
curious about this. I would have thought that beyond Posix compliance
that the diversity amongst operating systems was such that you would
be hard-pressed finding a common another set of features/interfaces
that could unambiguously define a family called "Unix".
>> >Formal standards are nice for justifying a purchase request to the bean
>> >counters.
>
>> But you are doing a diservice to yourself and your bean-counters if
>> the formal standards specified in your purchase request don't service
>> a real pratical need.
>
>Why, of course! That's why I would never specify POSIX.
Fine. But suppose you want "Unix" functionality? My point is that
however incomplete Posix may currently be, it still the best available
standard.
>No, my premise is that POSIX is incomplete. In response to your implication
>that formal standards are by definition complete what can I say but that
>it isn't so.
Of course. There was no such implication from me.
>
>> --- how about some specifics (and remember you've already been wrong
>> about crontab :-).
>
>OK. Port CNEWS to OpenVMS. Provide justification for any modification to
>any file outside of the "conf" directory, as to why that modification would
>be necessary on a UNIX system. C News is one of the more portable programs
>I have worked on. Henry Spencer did an excellent job.
I agree that this would be a good test of the utility of the Posix
standards.
Carl J Appellof
> Then, could you show us a little program, which uses TCP/IP networking
> without violating POSIX?
Obviously, I can't, since TCP/IP networking doesn't appear anywhere
in the POSIX standard.
BUT, POSIX for OpenVMS V2.0 has integrated support for TCP/IP using
BSD-style sockets. Last year my group ported a rather large UNIX
client/server application to VMS using the POSIX for OpenVMS
environment. It wasn't trivial, but wasn't impossible either.
The vagaries of the old VAXC compiler caused more problems than the
POSIX interface. Now with ANSI-standard DECC, things are easier.
By the way, POSIX didn't help us with the Motif interface either.
--
Carl J. Appellof (c...@chmist.zso.dec.com)
Storage Management Group
POLYCENTER NetWorker Save and Restore
Digital Equipment Corporation
John Hascall
}pe...@nmti.com "Peter da Silva" writes:
}> Douglas Miller <dou...@brt.deakin.edu.au> wrote:
}> > Posix *is* Unix by definition.
}> Um, no. By definition UNIX is "Spec 1170". Posix is a series of standards
}> that each define the API for a subset of UNIX. Posix is a good thing, but
}> it doesn't really go far enough.
}Well Digital state that VMS complies approx 85% with the (draft?) Spec1170
}document at the moment. How does this compare with the proprietary UNIXes ?
According to a table in the July '94 _Digital Unix News_:
DEC OSF/1 HP-UX Solaris IBM AIX
V2.0 V9.0 V2.3 V3.2
% of Spec 1170
API present 99.5% 56.6% 95.6% 61.1%
Spec 1170
conformance 92% 48% 85% 45%
John
--
John Hascall ``An ill-chosen word is the fool's messenger.''
Systems Software Engineer
Project Vincent
Iowa State University Computation Center + Ames, IA 50011 + 515/294-9551
Peter da Silva
Roger Barnett <Ro...@natron.demon.co.uk> wrote:
> In article <id.4LY...@nmti.com> pe...@nmti.com "Peter da Silva" writes:
> > In article <325f4m$a...@sol.ccs.deakin.edu.au>,
> > Douglas Miller <dou...@brt.deakin.edu.au> wrote:
> > > Posix *is* Unix by definition.
> > Um, no. By definition UNIX is "Spec 1170". Posix is a series of standards
> > that each define the API for a subset of UNIX. Posix is a good thing, but
> > it doesn't really go far enough.
> Well Digital state that VMS complies approx 85% with the (draft?) Spec1170
> document at the moment. How does this compare with the proprietary UNIXes ?
So far I haven't heard any vendor issue that sort of premature announcement...
why would they? They don't need to force the issue.
But ignoring your terminology at this point, Spec 1170 is in the process of
being defined as a reflection of these systems. It's unlikely that they will
spec something they can't easily match.
(as for "85% compliance", that would at best mean that VMS was 85% of UNIX)
What is the whole point in defining VMS as UNIX anyway? It doesn't use the
basic UNIX model internally, so native software is going to be ill served
by avoiding the VMS interface. If you want VMS, it's something in the VMS
side of things you're buying it for. If you want a UNIX system, you'll buy
a UNIX system and avoid all the emulation overhead. Is this just to make
it easier to justify things to bean counters?
as a line item on a
J Lee Jaap
|>John Hascall <jo...@iastate.edu> writes:
|>
|>>which isn't all that different from on a Ultrix machine: :)
|>>
|>># crontab -e
|>>crontab: Command not found.
|>>
|>
|>which is why ultrix <> unix or at least ultrix < unix
But I thought that Unix was the same everywhere, the grand unified OS!?
<* big ingenuous open-mouth stunned expression *>
No. There's Solaris 1, Solaris 2, Ultrix, OSF/1, AIX, IRIX, A/UX,
Unicos, HP/UX, Linux, BSD, System 5, ... ad nauseam. They _all_
differ, sometimes unexpectedly for programers and users. On one of
the various flavors of Unix:
% uname
uname: Command not found
% whereis uname
whereis: Command not found
Does ls automatically output in columns to a terminal?
What options does stty take?
Does sh understand 'set -- blah'?
If the /usr partition is corrupt, is single-user mode functional?
Does csh glob, or do individual programs?
What's the standard? Posix? OSF? System V? BSD?
Don't get me wrong. Some Unixes are good OSes; VMS is also a good OS;
so is MacOS/Finder. I use various Unixes, and like the tinker-toy
method of building new commands (pipes), but get frustrated at the
diffs. I have used VMS and liked the 'fullness'. MacOS/Finder gives
a 'complete' (if it's not in a menu, it's not available), recognize-
not-remember environment to the user. If I wrote an OS, I'd combine
features of them all. Pipes are nice, but so is being able to
truncate commands to an unambiguous prefix. (Yah, the second is
obviously a shell/CI issue, but the whole picture (sysadmin,
programming, user) is important.)
Cheers.
--
J Lee Jaap <J.L....@LaRC.NASA.Gov> +1 804/864-2148
employed by, not necessarily speaking for, AS&M Inc,
at NASA LaRC, Hampton VA 23681-0001
Peter da Silva
Douglas Miller <dou...@brt.deakin.edu.au> wrote:
> In Article <id.NLY...@nmti.com>
> pe...@nmti.com (Peter da Silva) writes:
> >> >POSIX is a useful spec for writing portable applications,
> >> >if your application doesn't have to know much about its environment.
> >> Kind of follows from the definition of "portable".
> >No.
> Yes. I'm saying you have made a tautology.
Only if you think "portable" is a binary condition: that a program is
either portable or it isn't.
> >> >Linux is "unix". MINIX is "unix". Neither of these have line 1 of AT&T
> >> >code in them.
> >> Yes I know. But are they Unix?
> >Yes.
> Why do you think this (I'm curious)?
OK, abandoning ALL practical definitions (because they all come down to what
software you can port to the system) and taking a philosophical stance.
UNIX is:
1. A common hierarchical name space for all programs on the system
that contains all files and devices, and (with few exceptions)
is the mechanism for controlling access to objects.
2. A standard opaque object, the file descriptor, that acts as the
common intermediary for all transactions.
3. A small common set of operations that are available on that
object. Version 7 has 35 system calls, and while there's a lot
more than that in System V a large subset of these 35 are still
there and cover most of the things you do.
On a UNIX system, you can (with appropriate privilege) access any object
in the system by calling "open", "read", "write", and so on.
ANY object. Not just objects that have been exposed to the POSIX subsystem.
That means that if I "su" and edit /etc/passwd I've done something... I
don't have to go back and dink with SYSUAF.DAT or whatever it's called.
Now there are more and more exceptions to this, and the more there are the
less a system is UNIX. The worst case that can still be called UNIX is
System V, with the abominable IPC mechanisms and ridiculous extra name
spaces for shared memory, streams, and such.
Berkeley network domain sockets are another case, but they're acceptable
since they operate outside the domain of the system. UNIX domain sockets
are acceptably UNIXy.
In other ways, UNIX systems are getting closer to the ideal. Just look at
the interesting file systems out there like ALEX.
The reason VMS isn't UNIX is that there are programs you have to have to
run the system that don't operate on the UNIX interfaces. It contains an
increasingly good UNIX emulator, and one might even be able to say that
the POSIX subsystem is UNIX one of these days, but VMS itself does not use
the UNIX model.
This doesn't necessarily mean that it's bad, in and of itself, just that
it's not UNIX.
> Fine. But suppose you want "Unix" functionality?
Then I specify "Unix".
Peter da Silva
Douglas Miller <dou...@brt.deakin.edu.au> wrote:
> OK, terminology problem --- I don't think "violating" is correct. If
> you had instead said that "Posix doesn't provide all the interfaces
> you will need for every application", then I would have agreed.
OK. I'm used to working with standards where a program that uses an extension
is considered to be violating that standard. If you don't like the word
"violating" I'll be happy to skip it.
> Well in VMS the pty's are under /dev as you would expect, and there is
> socket support for TCP (and IP and UDP), but beyond that I don't know
> enough Posix to comment in detail. You've obviously studied Posix in
> some depth --- please tell us the answers.
The answer is that, so far as I know, POSIX doesn't provide these interfaces.
It doesn't address system management issues.
Mark H. Wood
[deletia]
>> > Does it run SCO binaries? Unixware binaries? Solaris binaries?
>> > Does it satisfy the SVID?
>> > Does it support the BSD API (native sockets, for example).
>
>> Now we are really in looking-glass land. Vendor-independent
>> interfaces are "useless", and proprietary interfaces are worthwhile?!
>
> The BSD API is a proprietary interface? Who owns it?
The Trustees of the University of California? Maybe that's why their copyright
is all over it?
--
Mark H. Wood, Lead Systems Programmer +1 317 274 0749 [@disclaimer@]
Internet: MW...@INDYVAX.IUPUI.EDU BITNET: MWOOD@INDYVAX
"I live the greatest adventure one could ever wish." - a tosc
obe...@icaen.llnl.gov
imh...@indyvax.iupui.edu (Mark H. Wood) writes:
>In article <id.YLY...@nmti.com>, pe...@nmti.com (Peter da Silva) writes:
>> The BSD API is a proprietary interface? Who owns it?
>
>The Trustees of the University of California? Maybe that's why their copyright
>is all over it?
Not to be picky... (OK, I'm being picky!)
It's the Regents of the University of California. I ought to know. They pay my
salary (from money the DOE pays to them).
R. Kevin Oberman
Energy Sciences Network (ESnet)
National Energy Research Supercomputer Center (NERSC)
Lawrence Livermore National Laboratory (LLNL)
Internet: kobe...@llnl.gov +1 510-422-6955
Douglas Miller
>Roger Barnett <Ro...@natron.demon.co.uk> wrote:
>> In article <id.4LY...@nmti.com> pe...@nmti.com "Peter da Silva" writes:
>> > In article <325f4m$a...@sol.ccs.deakin.edu.au>,
>> > Douglas Miller <dou...@brt.deakin.edu.au> wrote:
>> > > Posix *is* Unix by definition.
>> > Um, no. By definition UNIX is "Spec 1170". Posix is a series of standards
>> > that each define the API for a subset of UNIX. Posix is a good thing, but
>> > it doesn't really go far enough.
Last I heard of spec 1170 it was in its infancy (i.e., no actual
conformant O/'s existed) and was not a functional spec (there was a
clause that also required a conformant O/S to also be in some way a
descendant of original "Unix"). If this is changed and spec 1170 is
now the leading candidate for a formal definition of Unix, then I will
happily stand corrected.
>> Well Digital state that VMS complies approx 85% with the (draft?) Spec1170
>> document at the moment. How does this compare with the proprietary UNIXes ?
>What is the whole point in defining VMS as UNIX anyway?
What is the point in defining Solaris/HP-UX/Windows-NT as UNIX?
The point is that a formal functional definition of "Unix" (whether it
be Posix, spec 1170 or whatever) provides the best currently available
standard for application portability, i.e., for the first time the
prospect of being able to mix and match conformant O/S'es and
applications is starting to look like a real possibility.
>It doesn't use the basic UNIX model internally,
You said "By definition UNIX is 'spec 1170'", which obviously doesn't
include an internal model --- basic or otherwise.
>so native software
I assume you mean "internal" rather than "native" (the emulation status
of the software hardly seems relevant).
> is going to be ill served by avoiding the VMS interface.
You mean the parts of the VMS interface beyond the basic Unix
conformance? Why would internal software need to avoid these --- by
definition it is non-portable anyway.
>If you want VMS, it's something in the VMS side of things you're
> buying it for.
I can't be sure what point you are making with this truism. Are you
buying into the old myth that "unix" conformance is going to make all
O/S's identical?
If you want VMS, it's because VMS has the features that best meet your
requirements (which may include "Unix" compliance).
> If you want a UNIX system, you'll buy a UNIX system
VMS for example.
>and avoid all the emulation overhead.
What emulation overhead? You mean in VMS (so VMS is non-Unix now?)?
There is no emulation in VMS Posix. You have a choice of two CLIs:
DCL and Posix (which is Posix.2 compliant); they run independently,
neither emulates the other. Similarly Posix.1 system calls execute
natively, just like VMS System Services.
Terry Kennedy, Operations Mgr.
[Long discussion deleted]
> There is no emulation in VMS Posix. You have a choice of two CLIs:
> DCL and Posix (which is Posix.2 compliant); they run independently,
> neither emulates the other. Similarly Posix.1 system calls execute
> natively, just like VMS System Services.
I think there are two problems with VMS Posix: First, it is specifications
for a subset of the "standard Unix-like operating system", hashed out by a
standards committee. It doesn't cover a number of areas of interest to folks
working on portable software. As an example, try running anything that uses
"Configure" on VMS Posix. Trying this with Perl's Configure would be partic-
ularly edifying. Note that this works on many platforms, and can be made to
work with trivial edits on many more. Now note the things it trips over on
VMS Posix, the fact that they're not covered by the Posix spec, and that they
are common enough that Configure works "everywhere" else.
Second, the Posix spec doesn't include any requirements for execution speed.
There are common assumptions about the cheapness of certain operations when
done on a Unix-like system that don't hold true on VMS. For example, process
creation speed, pipe speed, and the efficiency of certain disk operations.
While it's true that VMS has other methods for doing these things (and it can
be argued that the VMS methods may be "better"), there isn't a clean way to
map the Unix methods onto the VMS ones.
There are valid reasons to use VMS for projects - I just don't think that
Posix compliance is a valid technical one. Sure, it is useful if you have a
few things that are easier to port than rewrite which aren't crucial to the
project, or if you need to cite Posix compliance for political reasons to
win a bid (assuming you can go and do the project successfully with the native
VMS features).
Terry Kennedy Operations Manager, Academic Computing
te...@spcvxa.spc.edu St. Peter's College, Jersey City, NJ USA
+1 201 915 9381 (voice) +1 201 435-3662 (FAX)
Peter da Silva
Mark H. Wood <imh...@indyvax.iupui.edu> wrote:
> In article <id.YLY...@nmti.com>, pe...@nmti.com (Peter da Silva) writes:
> > The BSD API is a proprietary interface? Who owns it?
> The Trustees of the University of California?
They've all but given it up to the public domain. The Berkeley copyright notice
disclaims virtually all intellectual property rights, retaining only the right
to be given credit for the work. They have no control over it whatsoever.
> Maybe that's why their copyright is all over it?
I assume this means you read the notice, so what's your point?
Peter da Silva
Douglas Miller <dou...@brt.deakin.edu.au> wrote:
> conformant O/'s existed) and was not a functional spec (there was a
> clause that also required a conformant O/S to also be in some way a
> descendant of original "Unix").
There is no such clause. There is a clause grandfathering in certain such
OS-es, but that's a different matter.
> If this is changed and spec 1170 is
> now the leading candidate for a formal definition of Unix, then I will
> happily stand corrected.
>
> >> Well Digital state that VMS complies approx 85% with the (draft?) Spec1170
> >> document at the moment. How does this compare with the proprietary UNIXes ?
>
> >What is the whole point in defining VMS as UNIX anyway?
>
> What is the point in defining Solaris/HP-UX/Windows-NT as UNIX?
Solaris and HPUX use the underlying "Unix" programming model as their native
API. That means that a program written to assume those portions of the "Unix"
API that are not specified in POSIX will run on them. It also means that
coding for the "Unix" API is the most efficient way of working on those systems.
The point of labelling Windows-NT as "Unix" is a cynical marketing move to
confuse potential customers who will proceed to be frustrated by the crippled
compatibility box. Comparing OpenVMS (which I consider a good thing, by the
way) to NT does not reflect any credit on OpenVMS.
> The point is that a formal functional definition of "Unix" (whether it
> be Posix, spec 1170 or whatever) provides the best currently available
> standard for application portability, i.e., for the first time the
> prospect of being able to mix and match conformant O/S'es and
> applications is starting to look like a real possibility.
Replacing the text from "Unix" through the end of the following parenthesised
phrase with "Posix", I would agree completely. Posix *does* provide the best
currently available standard for application portability. Unix is a family
of operating systems from which Posix has subsetted those interfaces that do
not refer to system administration and related issues. For a good many
applications that is enough.
> >It doesn't use the basic UNIX model internally,
> You said "By definition UNIX is 'spec 1170'", which obviously doesn't
> include an internal model --- basic or otherwise.
>
> >so native software
>
> I assume you mean "internal" rather than "native" (the emulation status
> of the software hardly seems relevant).
Only if you consider performance irrelevant. For example, programs on VMS
will operate more efficiently if they avoid creating lots of processes,
while on "Unix" there's little penalty for doing that sort of thing. That's
an aspect of "Unix" that Posix doesn't specify (though I understand OpenVMS
is a lot better than older VMS systems in that respect).
Also, by using the POSIX interface you lose the ability to take advantage
of things like RMS, which are the reasons often given for picking VMS in the
first place. About the only benefit the POSIX environment on VMS has over
UNIX is clustering.
If you want VMS, use VMS. I don't say there's not reasons for doing so. If
you want to run Posix software on OpenVMS, go for it. But if you expect any
random UNIX package to work on OpenVMS as a result you'll be disappointed.
(and before you go on about UNIX variants and portability problems: I've got
a wide variety of UNIX systems here and I don't *see* the horrible portability
problems except where there's obvious differences in hardware architecture.
Running code written on a VAX-class machine like a Sparc, on a PDP-11 class
machine like a 286, is going to take some work. That it's possible at all
under UNIX is a testimonial to portability... how much VMS code runs at all
on RSX-11 (I've ported stuff both ways, so I'll answer that... damned little
and all of that's in Fortran))
> I can't be sure what point you are making with this truism. Are you
> buying into the old myth that "unix" conformance is going to make all
> O/S's identical?
Heavens no. What I'm saying is that the only people who are going to be
buying a VMS system if you call it "Unix" instead of "Posix" are people
who don't care if it's VMS or not. I'm saying that calling it "Unix" instead
of "Posix" is confusing the issue at best, and a cynical marketing tactic
like Bill gates declaring that NT is UNIX at the worst.
I'm trying not to make a value judgement here. I've repeated over and over
again that Posix is a good thing, for example, because it is. OpenVMS is
a good thing. But it's still not "Unix".
> If you want VMS, it's because VMS has the features that best meet your
> requirements (which may include "Unix" compliance).
If your requirements are for "Posix" compliance and you specify "Unix"
compliance, then your spec is broken. If your requirements are for "Posix
plus this feature from Unix and that feature from VMS" and OpenVMS provides
that feature, go for it.
> What emulation overhead? You mean in VMS (so VMS is non-Unix now?)?
VMS isn't "non-UNIX now", it's "non-UNIX".
> There is no emulation in VMS Posix. You have a choice of two CLIs:
> DCL and Posix (which is Posix.2 compliant); they run independently,
> neither emulates the other. Similarly Posix.1 system calls execute
> natively, just like VMS System Services.
Uh-huh. We set up some OpenVMS systems here to check on how close to "Unix"
they are, since we had been using VMS and Sparcs.
We ported our stuff to OSF/1 instead...
OpenVMS provides a good POSIX subsystem, but that's as far as it goes.
Mark H. Wood
> In article <1994Aug12.112947.7410@ivax>,
> Mark H. Wood <imh...@indyvax.iupui.edu> wrote:
>> In article <id.YLY...@nmti.com>, pe...@nmti.com (Peter da Silva) writes:
>> > The BSD API is a proprietary interface? Who owns it?
>
>> The Trustees of the University of California?
>
> They've all but given it up to the public domain. The Berkeley copyright notice
> disclaims virtually all intellectual property rights, retaining only the right
> to be given credit for the work. They have no control over it whatsoever.
>
>> Maybe that's why their copyright is all over it?
>
> I assume this means you read the notice, so what's your point?
My point is that they *own* it. It is the property of the Regents, and they
can decide what other people will be allowed to do with it. That they decide
to let everybody use it without fee is immaterial to whether it is proprietary.
Allowing free use is very, very different from placing a work in the public
domain.
Peter da Silva
Mark H. Wood <imh...@indyvax.iupui.edu> wrote:
> My point is that they *own* it. It is the property of the Regents, and they
> can decide what other people will be allowed to do with it.
I don't think they could withdraw the current sources at this point.
> Allowing free use is very, very different from placing a work in the public
> domain.
So anything not in the public domain is an evil proprietary interface? I
guess that includes POSIX. Why... that includes all the ANSI and ISO
standards as well. Gee, I guess either proprietary isn't evil or you're
misusing the word.
Mark H. Wood
> In article <1994Aug16.105300.7508@ivax>,
> Mark H. Wood <imh...@indyvax.iupui.edu> wrote:
>> My point is that they *own* it. It is the property of the Regents, and they
>> can decide what other people will be allowed to do with it.
>
> I don't think they could withdraw the current sources at this point.
Nope. They could decide not to freely distribute updated sources some day, but
probably won't.
>> Allowing free use is very, very different from placing a work in the public
>> domain.
>
> So anything not in the public domain is an evil proprietary interface? I
> guess that includes POSIX. Why... that includes all the ANSI and ISO
> standards as well.
Evil? Not necessarily. Proprietary? Yes. To say that a thing is proprietary
is to say that someone owns it, that it is property. Despite the mantras that
many have been chanting for so long, Unix is proprietary: the name and
description belong to X/Open, and the implementations belong to Sun, Novell,
DEC, HP.... Anyone who thinks this is not so, should try distributing one of
them without first obtaining a license to do so.
> Gee, I guess either proprietary isn't evil or you're
^^^^^^^^^^^^^^^^^^^^^^
> misusing the word.
Bingo! Lots'n'lots of people *are* misusing the word.
Hmmm, when you think about it, something that is not in the public domain ought
to be in some *private* domain, i.e. somebody owns it.
Peter da Silva
Mark H. Wood <imh...@indyvax.iupui.edu> wrote:
> Evil? Not necessarily. Proprietary? Yes. To say that a thing is proprietary
> is to say that someone owns it, that it is property.
According to one definition, yes. But this usage of the word is so broad as
to be meaningless. It is not in any way useful in debate, and is certainly
*not* the opposite of "open" which is how most people use it.
> Despite the mantras that
> many have been chanting for so long, Unix is proprietary:
In terms of "a single entity owns it", yes. In terms of "a single entity
controls it", no. The only thing that controls Unix is the market.
> the name and
> description belong to X/Open, and the implementations belong to Sun, Novell,
> DEC, HP.... Anyone who thinks this is not so, should try distributing one of
> them without first obtaining a license to do so.
If you send me your address I'll send you a free CDROM that you can in turn
distribute any way you want without paying anyone a fee or acquiring a
license. It's not *called* UNIX, for legal reasons, but by any practical
test it's a better implementation of the UNIX design than some of the
systems you named. It's certainly more UNIX than AIX is.
Mark H. Wood
> In article <1994Aug17.111743.7539@ivax>,
> Mark H. Wood <imh...@indyvax.iupui.edu> wrote:
>> Evil? Not necessarily. Proprietary? Yes. To say that a thing is proprietary
>> is to say that someone owns it, that it is property.
>
> According to one definition, yes. But this usage of the word is so broad as
> to be meaningless. It is not in any way useful in debate, and is certainly
> *not* the opposite of "open" which is how most people use it.
Reaaly. I thought it considerably narrower than the way in which I usually
hear it used, which is indeed as an antonym of "open". So, what do *you* think
"proprietary" means? What does your attorney think it means?
>> Despite the mantras that
>> many have been chanting for so long, Unix is proprietary:
>
> In terms of "a single entity owns it", yes. In terms of "a single entity
> controls it", no. The only thing that controls Unix is the market.
Nonsense. X/Open controls it. They decide what is Unix and what is not. The
only control that the market has, is that if X/Open doesn't follow the market
then the market won't value the XPG brandmark, or the Unix trademark.
>> the name and
>> description belong to X/Open, and the implementations belong to Sun, Novell,
>> DEC, HP.... Anyone who thinks this is not so, should try distributing one of
>> them without first obtaining a license to do so.
>
> If you send me your address I'll send you a free CDROM that you can in turn
> distribute any way you want without paying anyone a fee or acquiring a
> license. It's not *called* UNIX, for legal reasons, but by any practical
> test it's a better implementation of the UNIX design than some of the
> systems you named. It's certainly more UNIX than AIX is.
Thank you for your kind offer. I'm writing this from a Linux system already.
Linux is distributed under license, and happens to be free. It is the property
of Linus Torvalds, who can decide not to give it away any more if he wishes.
If you tried to *sell* Linux, you'd be in violation of the license and could
theoretically be sued, which would be an interesting test of the implicit-
license concept.
I haven't seen the FreeBSD terms, but I'd be surprised to find they are much
different.
Wayne Scott
One UNIX difference that has driven me nuts as I've moved from
AIX to OSF/1 to SunOS is how the "ping" command works.
On AIX and OSF/1 it does something like this:
============================================================
isp$ ping navaho
PING navaho.cc.bellcore.com (128.96.96.216): 56 data bytes
64 bytes from 128.96.96.216: icmp_seq=0 ttl=252 time=66 ms
64 bytes from 128.96.96.216: icmp_seq=1 ttl=252 time=4 ms
64 bytes from 128.96.96.216: icmp_seq=2 ttl=252 time=4 ms
64 bytes from 128.96.96.216: icmp_seq=3 ttl=252 time=3 ms
^C
----navaho.cc.bellcore.com PING Statistics----
4 packets transmitted, 4 packets received, 0% packet loss
round-trip (ms) min/avg/max = 3/19/66 ms
isp$ echo $?
0
isp$
===========================================================
On AIX, it returns a 1 if it was not successful.
On OSF/1 it ALWAYS returns 0.
I have to do a popen() and scan the output for the % packet loss.
On SunOS it tells me:
[wws@microjet] ping 128.96.96.216
128.96.96.216 is alive
[wws@microjet]
It's a real nuisance trying to port "simple subroutines"
that check the state of target systems.
Wayne
--
----------------------------------------------
w...@cc.bellcore.com
I'm just a soul whose intentions are good,
Oh Lord, please don't let me be misunderstood.
Dan Pop
>In article <id.DS5...@nmti.com>, pe...@nmti.com (Peter da Silva) writes:
>> In article <1994Aug16.105300.7508@ivax>,
>> Mark H. Wood <imh...@indyvax.iupui.edu> wrote:
>>> My point is that they *own* it. It is the property of the Regents, and they
>>> can decide what other people will be allowed to do with it.
>>
>> I don't think they could withdraw the current sources at this point.
>
>Nope. They could decide not to freely distribute updated sources some day, but
>probably won't.
And who is supposed to produce the updated sources, since the BSD Unix
project is history today?
>
>Evil? Not necessarily. Proprietary? Yes. To say that a thing is proprietary
>is to say that someone owns it, that it is property. Despite the mantras that
>many have been chanting for so long, Unix is proprietary: the name and
>description belong to X/Open, and the implementations belong to Sun, Novell,
>DEC, HP.... Anyone who thinks this is not so, should try distributing one of
>them without first obtaining a license to do so.
>
What about Linux, NetBSD and FreeBSD? Or aren't they Unices?
Peter da Silva
Mark H. Wood <mw...@indyvax.iupui.edu> wrote:
> Reaaly. I thought it considerably narrower than the way in which I usually
> hear it used, which is indeed as an antonym of "open". So, what do *you* think
> "proprietary" means? What does your attorney think it means?
You're a lawyer? I might have figured. A lawyer's someone who gives you
information that's completely accurate and completely useless.
> > In terms of "a single entity owns it", yes. In terms of "a single entity
> > controls it", no. The only thing that controls Unix is the market.
> Nonsense. X/Open controls it. They decide what is Unix and what is not. The
> only control that the market has, is that if X/Open doesn't follow the market
> then the market won't value the XPG brandmark, or the Unix trademark.
And that's the only meaningful control.
> Thank you for your kind offer. I'm writing this from a Linux system already.
Actually, FreeBSD.
> Linux is distributed under license, and happens to be free. It is the property
> of Linus Torvalds, who can decide not to give it away any more if he wishes.
Yes, and then new code written by him would not be available. Code already
licensed under the GPL would remain available.
> If you tried to *sell* Linux, you'd be in violation of the license and could
> theoretically be sued, which would be an interesting test of the implicit-
> license concept.
Not so. You can sell Linux all you want so long as you sell it with source
and don't try and restrict distribution. That makes it kind of hard to make
money, but it's possible.
In any case, I guess you're not a lawyer after all. Lawyers tend to read
contracts before talking about them.
> I haven't seen the FreeBSD terms, but I'd be surprised to find they are much
> different.
They're *very* different. It is distributed under the Berkeley license. It
deliberately has NO GPL-protected code in the kernel, and the GPL source is
isolated in a specific part of the source tree where GCC and so on live.
You are specifically allowed to take it and sell it commercially.
In fact one company, Berkeley Software Design, Incorporated, is doing just that
with the original 4.4 BSD-Lite code. I would say that in practical terms, that
product is probably the best commercial UNIX available on the market today.
Even if it's not *called* UNIX.
Peter da Silva
> If Linus Torvalds decides to turn it into a commercial product, he'll
> have rewrite it from scratch, according to the terms of the GPL.
No he won't. The GPL doesn't restrain the author of a piece of code from
doing anything he wants with it. He can't withdraw the existing code, but
he can release future versions under any license he wants subject to the
licenses in any components he's using that other people provided.
Ghostscript in fact was just changed to commercial in just this way. gs2.6
and prior versions remain GPL. gs3.0 and subsequent versions are pretty
much shareware. gs3.0 has significant improvements, though I prefer the
epson driver in gs2.6.
Terry R. Friedrichsen
both VMS and Unix, and have no particular axe to grind about either, though I
much prefer Unix.
1) The VMS people are now desperately trying to convince the world that VMS
is, in fact, Unix. Note that the reverse is *not* true; the Unix people are
not trying to convince anybody that Unix is, in fact, VMS.
Take a rubber glove, poke holes in its fingertips, fill it with milk, and
attach it to a bull. A certain amount of emulation is taking place, but
you ain't no way turned it into a cow ...
2) The VMS people are now desperately trying to convince the world that
Unix is proprietary, too.
This strikes me as an interesting adult variation of the "oh, yeah? well,
you're one, too!" arguments I recall from childhood. Even assuming the
argument to be correct (which it isn't, of course; you can get full and com-
plete source code to at least four implementations of Unix, and nearly all
the sources to a fifth), how does this in any way established the superiority
of VMS?
Please don't misconstrue this as any denigration of the original posters;
all I'm saying is that the VMS arguments have clearly failed to take the
high moral ground ...
Terry R. Friedrichsen
te...@venus.sunquest.com (Internet)
uunet!sunquest!terry (Usenet)
te...@sds.sdsc.edu (alternate address; I live in Tucson)
John Wilson
Terry R. Friedrichsen <te...@venus.sunquest.com> wrote:
>you can get full and com-
>plete source code to at least four implementations of Unix, and nearly all
>the sources to a fifth
I've seen this before, and I must say it strikes me as a WEIRD way to make
the distinction between "proprietary" and "open". Just this morning, a
friend and I picked up some used DEC stuff we'd bid on, including an old
VAX-11/750. In the big boxes of goodies that came with it (don't they
always) was what looks like complete fiche of the sources to VMS 3.0
through 3.3. Well then, I guess as far as we're concerned, those particular
versions of VMS just made the transition from "proprietary" to "open".
Yeah right! Well then I guess RT-11 must not be proprietary either, since
it always comes with sources (you need them to SYSGEN). But MS-DOS *is*
proprietary since it doesn't, even though there are several completely
independent re-implementations of its API, some of which are free. Strange!
I think what it boils down to is (a) whether a company is poised to sue
the living shit out of anyone who uses the same interface as they do
(definitely proprietary if so, like when Apple sued Franklin into orbit),
and (b) if not, whether anyone actually does (how "open" is a standard that
hardly anyone uses? ITS has always been an open system, but a lot of good
that's done; how many manufacturers provide MLDEV clients or servers?).
John Wilson
Peter da Silva
Terry C Shannon <sha...@world.std.com> wrote:
> I think the OpenVMSers are trying to counter the perception that OpenVMS
> is closed and xenophobic.
I don't recall claiming that OpenVMS was closed or xenophobic. Posix is
great, and if you can run the stuff you need on OpenVMS under the Posix
subsystem, that's great too.
> >2) The VMS people are now desperately trying to convince the world that
> >Unix is proprietary, too.
> If "proprietary" means "you pay for it" then Unix sure as heck is
> proprietary.
No, proprietary means "you buy it, you stuck with us".
If we had to run VMS to use the Alpha, we wouldn't be using the Alpha, because
we don't know how well DEC will be doing a few years from now and we've been
burned on dead product lines before. But OSF/1 software will move easily
enough to Sun or HP or whoever, maybe even FreeBSD. Meanwhile the TPU debacle
demonstrates how portable VMS software is...
Fred Kleinsorge
>1) The VMS people are now desperately trying to convince the world
that VMS
>is, in fact, Unix. Note that the reverse is *not* true; the Unix
people are
>not trying to convince anybody that Unix is, in fact, VMS.
>
I'll walk around the floor and ask... but I don't think that anyone
here thinks that VMS is Unix. There *have* been some posters in
this and other newsgroups that have used the argument in trying to
make their points - but I don't think that this is taken serious
by anyone except for purposes of rhetoric.
>2) The VMS people are now desperately trying to convince the world
that
>Unix is proprietary, too.
>
I think more often than not, the argument lies in just what "open"
means. Some people believe Open = Unix. And that Unix = Unix.
Others know the subtle flaws in practice, that a *useful* Unix is
generally proprietary in many of the features that make it useful.
Another argument is that MS-DOS and Windows is far more "open"
if the whole point of being "open" is to remove the dependence on a
single vendor for hardware. Still others believe that if you can
support a core set of standards, then properly written code can
be independent of both O/S *and* hardware (at least at the source
level).
--
+--------------------------------------+
Fred Kleinsorge | All opinions expressed are mine, and |
klein...@star.enet.dec.com | may not reflect those of my employer |
+--------------------------------------+
fair...@slacvx.slac.stanford.edu
[...]
> If we had to run VMS to use the Alpha, we wouldn't be using the Alpha, because
> we don't know how well DEC will be doing a few years from now and we've been
> burned on dead product lines before. But OSF/1 software will move easily
> enough to Sun or HP or whoever, maybe even FreeBSD. Meanwhile the TPU debacle
Well, perhaps the volume of customer demand needs to be raised a
few notches (gee, I thought everyone who wants unix would naturally
be happy with emacs :-). I think "debacle" sounds a bit strong.
You may have missed my recent post to comp.os.vms: this subject
(dropping support for TPU on Ultrix, and the lack of a port to OSF/1
and other unixes) came up there within the last two weeks, and has
been raised on vmsnet.tpu several times over the last year or more.
Brian McCarthy, a former _TPU_ developer (but still with Digital :-)
explained what was required to port to Ultrix (the Ultrix people had
to add some RMS-like hooks for TPU). The fact is that _currently_
no development is being done on VMS' TPU, let alone OSF/1 (or heaven
forbid, WNT :-). However, from hints Brian gave, there are people
inside Digital that _still_ want to do the port (to OSF/1 at least),
and the possiblity of that happening isn't precisely zero. :-)
My point is, I've heard a _few_ people complain about the lack
of TPU on (generic) unix, a couple complain about dropping support
for it on Ultrix, but I haven't seen a ground-swell demand that
Digital do the port. I expect Digital thinks that unix users want
vi and emacs and couldn't care less about TPU. Are they wrong?
BTW, I've also heard many complaints that 3rd party version of TPU
(NuTPU) isn't an adequate substitute (you can't really port your
extensions to EVE to NuTPU because of bugs and missing
functionallity). If you consider not having TPU on OSF/1, have you
complained LOUDLY to Digital about (so they think there might me a
market for it), or did you just sit back and grumble?
-Ken
--
Dr. Kenneth H. Fairfield | Internet: Fair...@Slac.Stanford.Edu
SLAC, P.O.Box 4349, MS 98 | DECnet: 45537::FAIRFIELD (45537=SLACVX)
Stanford, CA 94309 | Voice: (415) 926-2924 FAX: (415) 926-4335
----------------------------------------------------------------------------
These opinions are mine, not SLAC's, Stanford's, nor the DOE's...
Paul Cote
>1) The VMS people are now desperately trying to convince the world that VMS
>is, in fact, Unix. Note that the reverse is *not* true; the Unix people are
>not trying to convince anybody that Unix is, in fact, VMS.
you'll never hear a VMS software engineer make that claim.
Ross Alexander
Probably not, considering the relative scarcity of these beasts ;).
regards,
Ross
not a VMS software engineer
--
Ross Alexander VE6PDQ r...@cs.athabascau.ca,
(403) 675 6311 r...@auwow.cs.athabascau.ca
Television is chewing gum for the eyes. -- Frank Lloyd Wright
Terry C Shannon
>both VMS and Unix, and have no particular axe to grind about either, though I
>much prefer Unix.
>
>1) The VMS people are now desperately trying to convince the world that VMS
>is, in fact, Unix.
Although OpenVMS, unlike some Unix implementations, has achieved XPG4
compliance, I sincerely doubt that anyone in a postion of authority at
DEC is seriously trying to convince people that OpenVMS is Unix. Even
Jesse Lipcon, who was quoted some time ago as saying, "Gee, maybe we
oughta call it [OpenVMS} Unix," was being facetious.
I think the OpenVMSers are trying to counter the perception that OpenVMS
is closed and xenophobic.
>2) The VMS people are now desperately trying to convince the world that
>Unix is proprietary, too.
If "proprietary" means "you pay for it" then Unix sure as heck is
proprietary. DEC doesn't give DEC OSF/1 away. Nor does HP give away
HP/UX. Or IBM, or Sun, etc. In this respect, Unix, OpenVMS, NT,
NetWare, OS/2 et al are equals.
>All I'm saying is that the VMS arguments have clearly failed to take the
>high moral ground ...
When you're mired in a marketing morass, it's difficult to take any sort
of high ground. But methinks some doth protest too much about morality.
It's about money and market share!
------------------------------------------------------------------------
Terry C. Shannon 135 Leland Farm Road
Senior Analyst Ashland, MA 01721
Illuminata 508-881-5563
sha...@illuminata.com