bobm@qip.UUCP (Bob Maccione) (09/11/90)
hey out there, did anybody go to QNX 90 and if so what happened? bob
frank@mnetor.UUCP (Frank Kolnick) (09/13/90)
In article <4480@qip.UUCP> bobm@qip.UUCP (Bob Maccione) writes: > >hey out there, did anybody go to QNX 90 and if so what happened? > >bob We (Basis Computer Systems Inc.) went, and we had a great time. The first night, Quantum rented the Museum of Civilization, which has an Omnimax projector, and we had a private screening of "The Dream is Alive". This was followed by a very nice dinner in the great hall, and a keynote address by Phil Lemmons (ex-Byte and ex-Personal Workstation editor). In fact, all of the meals were above average, and the entire event was quite sociable. On the technical side, the emphases were on the new Posix version of QNX, QNX Windows, and the Watcom compiler. I thought all of these were quite impressive, but you'll have to discount my strong bias towards the windows product, since it was designed and written by Basis. Quantum also re-iterated their commitment to customer support, in particular, continued support for all previous versions of QNX. There were 34 exhibitors this year (according to the official conference agenda), with a nice range of products. These include a toolkit for special effects (such as fountains) and other useful functions for QNX Windows,SCADA systems, compilers, communications products (TCP/IP, etc.), databases, SQL, etc. The free give aways were: a notepad, a pen, a QNX Windows t-shirt and a QNX Windows poster (artwork by Robert Tinney, btw). On the other hand, the weather prevented the annual balloon ride. All in all, a very well run conference. -- Frank Kolnick, Basis Computer Systems Inc. UUCP: {allegra, linus}!utzoo!mnetor!frank
frank@mnetor.UUCP (Frank Kolnick) (09/15/90)
copied from the QNX BBS with the author's permission...
==========================
qnx/QNX.90 #29, from stevehobbs, 2415 chars, Fri Sep 14 08:50:57 1990
--------------------------
To: All
From: Steve Hobbs
Date Sent: Thu 6 Sep 90 08:19
Subject: QNX'90
Day one was highlighted by Dan Dodge giving a summary of QNX 4.0
features. The POSIX release is scheduled to go beta this fall and
full release in the 1st quarter of '91. Pricing has not been decided.
The QNX 4.0 micro-kernel will be about 8K of code with only 16 entry
points and is only concerned with messaging! Dan commented that "this
would make porting to other processors easier". Don't hold your
breath for other versions though.
The rest of the OS is laid out modularly (more so than before). Only
the process manager (which includes timer management) is essential.
It implements three variants of pre-emptive scheduling. In addition
to the round-robin that exists today, tasks can choose not to be time
sliced with equal priority tasks, or can choose to age their priority
after receiving a time slice. Managers can choose to run at the
priority of their client process.
The BOOT process may now be run dynamicly under the NETBOOT process.
The OS in all its configurations is ROMable.
Device and file handling is even more transparent across the network
than before. Search paths are controlled by process rather than
processor. Environments (including open files and devices) are
inherited even when a task is created on a different node! The file
system implements real pipes and these can connect processes on
different nodes! The file system supports aliasing. The end result
is that it is possible to present a "single-system" image to the users
(both other processes and people).
The file system performance has been improved. Bill Flowers has done
a *marvelous* job. There are ideas in the new FSYS which have never
been done in any other OS. They were thinking of getting a patent on
one of the ideas. Bill was invited to write a paper to the ACM.
Instead they are going to "pull back the hood" and expose it for
everyone (especially us QNX developers) to see. There will be an
"inside QNX" book which will describe it all. Application or product
developers will be able to write their own file manager which adopts
a part of the name space.
Go nuts guys!!! This new release is going to be better than I had
been imagining or even hoping. It will be the OS of the 90's.
More later. Hold the questions 'til next week when the rest of the
boys get back home, then we can *really* fill the air.
==========================
qnx/QNX.90 #30, from stevehobbs, 771 chars, Fri Sep 14 08:52:52 1990
--------------------------
To: All
From: Steve Hobbs
Date Sent: Fri 7 Sep 90 07:15
Subject: QNX'90
The feature yesterday was on migration from 2.15 to 4.0. A 60 page
migration guide was released and a MIG24 program will be made available
to inspect and comment your C source code identifying problems and
recommended solutions. A set of cover functions will help for the short
term. The good news is that some *N*X code will port without change.
The new POSIX shell is a great improvement. You can actually have any
number of different shells provided you comment the first line of the
shell script with:
#! /bin/???
where ??? represents the shell program. There will be a QNX mini-shell
for compatibility with old shell scripts.
Gotta go to the final day's sessions. TTYL
==========================
qnx/QNX.90 #31, from stevehobbs, 1045 chars, Fri Sep 14 08:55:24 1990
--------------------------
To: All
From: Steve Hobbs
Date Sent: Sat 8 Sep 90 08:29
Subject: QNX'90
Yesterday, we saw a demo of Qwindows and its tools and a comparison to
an Xwindows application running on the same hardware.
The Xwindows demo needed most of the 8M RAM the machine had and was
slow; the Qwindows demo had 3 overlapped versions of the same demo
(PLAID) a windowed shell was brought up without disturbing the
response time of the other 3 windows too much (still faster than the
one Xwindow demo) and a TSK MEM showed a chunk of 5.1M still available
for applications. Absolutely awesome.
The toolkit for Qwindows is a "must get" if you will be doing a lot of
windowed applications. It allows you to build the windows from
components using a mouse rather than from within a program. Once your
window is built, you can test it and watch what events are produced
that your application would have to respond to. Quantum is thinking
about generating the .h needed as well as a boilerplate .c file that
can then be modified. This will be a real timesaver.
[Editor's note (i.e., me): the 'toolkit' Steve's referring to is really
a separate product called the interface editor. It allows you to construct
'base' windows, dialogs, etc. interactively, and then check them out by
examining the messages they send. You save the file and use it in apps
directly or as a template. All objects are position- and representation-
independent, i.e., you could re-arrange the objects in the window,
change the text to another language, etc. without modifying the
application at all. The app gets event messages which identify the
objects symbolically.]
==========================
qnx/QNX.90 #32, from stevehobbs, 1168 chars, Fri Sep 14 08:58:10 1990
--------------------------
To: All
From: Steve Hobbs
Date Sent: Sat 8 Sep 90 08:50
Subject: QNX'90
Thursday night was the "open season" for technical questions. The term
is a reference to the fact that 5 judges with loaded water pistols would
squirt either the technical panel or the questioner depending on the
outcome or nature of the question.
My first question was about that annoying habit of the command buffer
not working properly. The explanation is that control characters in the
buffer cause the shell to treat the command as a comment. While there
was no explation of how the control characters get into the buffer in
the first place, there was an assurance that the problem is gone under
4.0. Nobody got wet. Thanks to Dan Hildebrand for the answer.
My second question was about the accounting entries which occasionally
get messed up (when used in combination with ontty and a logoff) and WHO
reports noone on the node even though a number of shells are obviously
running. The answer was basicly "don't do that". I got wet as a
trouble maker.
There were other questions that stumped the techies and Bill Flowers won
the "wet Tshirt" contest. Good sports all around.
==========================
qnx/QNX.90 #33, from stevehobbs, 569 chars, Fri Sep 14 08:59:43 1990
--------------------------
To: All
From: Steve Hobbs
Date Sent: Sat 8 Sep 90 09:08
Subject: QNX'90
Friday morning saw a second round of technical questions. The most
interesting answer came from Dan Dodge who quickly and simply said
"Yes". The question was "Under 4.0, will I be able to transparent
networking across a dial-up line."
That means that downloads and uploads could be handled as a simple CP;
windows applications could be run at a site half-way around the world,
with the display work handled on the user's machine and the main
application being remote; ... Think about it!
--
Frank Kolnick,
Basis Computer Systems Inc.
UUCP: {allegra, linus}!utzoo!mnetor!frankfrank@mnetor.UUCP (Frank Kolnick) (09/15/90)
In article <5561@mnetor.UUCP> frank@mnetor.UUCP (Frank Kolnick) writes: >copied from the QNX BBS with the author's permission... Sorry about the indent/wrap towards the end of that message. It didn't look like that when I uploaded it :-) -- Frank Kolnick, Basis Computer Systems Inc. UUCP: {allegra, linus}!utzoo!mnetor!frank