Very Computer

Would you continue development under ODT 3.0 (would customers be able to
buy ODT 3.0 in 1998) or would you port it to OpenServer 5 (how much will
it change in 2+ years - the prototype does _not_ run under OS5) or wait
and see what comes out of the SCO/Unixware marriage and port to it?

--
----------------------------------------------------------


|  "Which ever it was that wasn't, or, the other way     |
|   around, or if not, the contrary."    J.P. Radley     |
----------------------------------------------------------

I read that ODT binaries should run under OS5 but in this case they do
not and I don't know why.  The binaries use Xwindows but that doesn't
seem to be the problem.  What I did notice is that there are a series
of binaries that need to be running in order for the HMI to come up but
under OS5, as soon as one of programs executes, it kills the previous
one.

Regardless, the project I work on is going to have to deal with the
main problem of what version of SCO is going to be available (sales and
support) in 1998.

Thanks for the reply.

Quote:> What if you are developing an application scheduled for general
> market release in 4Q 1998 and the current working prototype runs under
> ODT 3.0.  

> Would you continue development under ODT 3.0 (would customers be able to
> buy ODT 3.0 in 1998) or would you port it to OpenServer 5 (how much will
> it change in 2+ years - the prototype does _not_ run under OS5) or wait
> and see what comes out of the SCO/Unixware marriage and port to it?

        - ODT 3.0 COFF binaries
        - OpenServer 5 ELF binaries
        - UnixWare 2.x ELF binaries

So, your 3.0 binaries will still run Gemini.

I feel quite confident in stating that by Q4 1998 (almost 3 years from
now) we will no longer be selling ODT 3.0 :-).  By that time, we should
be, at least, phasing out OpenServer 5 and UnixWare 2 in favour of
Gemini.

So, you can take your pick as to binaries.  There will likely still be
a large installed base of ODT 3, OpenServer 5, and UnixWare 2.x in the
field at that time.  Heck, there will probably still be a large Xenix
installed base.  You will have to decide on profiles for your target
market, and decide whether those customers will have likely upgraded by
then.

You might also consider that the development systems in OpenServer 5
and UnixWare 2 are much better than the one in ODT 3.0.  You get a
better compiler, more APIs and much more compatibility between the 2.
Those devsys' are also much closer to what the Gemini devsys will
provide (please don't ask me for any more details on the Gemini devsys
... yet).

Hope this helps!

Larry

---
Larry Philps, Manager, ISV Engineering
SCO Canada Inc., 130 Bloor St. W., 10th Floor, Toronto, Canada M5S 1N5

=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
"The Internet treats censorship as a malfunction and just routes around it"
                                     - John Perry Barlow

I should have mentioned in my previous post "where" SCO has publically
stated stuff about Gemini ...

Check out the white paper "The Road to a "Merged" UNIX System at URL
http://www.sco.com/Products/Whitepapers/merged/merged1.htm.

You can also find a lot of "interesting" reading by going to the SCO
home page, http://www.sco.com/, and typing the work "gemini" into the
search engine.

Enjoy,

Larry

: Would you continue development under ODT 3.0 (would customers be able to
: buy ODT 3.0 in 1998) or would you port it to OpenServer 5 (how much will
: it change in 2+ years - the prototype does _not_ run under OS5) or wait
: and see what comes out of the SCO/Unixware marriage and port to it?

For what its worth - here is my $0.02:

From the looks of SCO's published roadmap, staying with ODT 3.0 would be
a problem.  From what I can tell, there will be several (more than two)
releases of SCO/Unixware before your planned delivery date.  That means
that I'd be willing to bet that they will not continue to support 4+ year
old software (though they may still sell it at that point).

I guess the bigger question, is what are you doing in your prototype that
causes it NOT to work on OS5?  From my understanding (and my visit to
SCO Forum '95) binaries that run under ODT 3 ARE supposed to run under
OS5.  What are you doing that is special - and have you tried compiling it
(on SCO 5) using the -a ods30 option?

jerry
--

Tactical Simulation Division   | http://www.arlut.utexas.edu/~jerry/main.html
Applied Research Laboratories of UT Austin | "Software is the difference
P.O. Box 8029, Austin, TX 78713-8029       |     between hardware and reality"

Oh, well...what else is new?

[ Tales of woe about an ODT3 app that "doesn't work" under OSR5" not quoted ]

Quote:>Thanks for the info.  I'd feel better, though, if the ODT 3.0 binaries
>worked under OS5 as claimed.  With this particular software, they
>don't. Some sort of system call/signal snafu, I assume.
>Oh, well...what else is new?

Sure, it's possible to write ODT3 binaries that don't work under
OpenServer.  I can write OpenServer binaries that run just fine on
my OpenServer box that don't run on your OpenServer box, too.  But
if your code is "playing by the rules", it should ahve no problems.

I have executed a boat-load of ODT3 compiled code under OSR5 and
have had no suprises about compatibility.  For example, I knew that
u386mon and memhog were groping private kernel data structures that
changed between the rleeases - hardly playing by the rules - so I
wasn't suprised then when didn't work.  [ They've since been ported. ]

I had exactly one app [ out of hundreds ] that became problematic in
the move.  It turns out it was tripping over a bug, which I then nailed
down, worked around,  and reproted to SCO.   Others have had different
success rates, for they've used different apps.

You need to have a programmer type analyze and understand the failure.
If you have an ODT3 app that doesn't run under OSR5, it probably won't
run on Linux, Unixware, Solaris, or anything else that claims to run
SCO binaries either.

--
Robert Lipe

Be it known that the application involved here is _not_ a commercial
off-the-shelf product.

Still, I read the brochure on OS5 and it said that binaries that run
under ODT 3.0 will (should?) run under OS5.  In my case, not so.  The
lesson here is to verify your apps will run under OS5 before upgrading.

I had the same type of problem several years ago with with an app that
ran under DOS 6.0 but not DOS 6.2.

: Would you continue development under ODT 3.0 (would customers be able to
: buy ODT 3.0 in 1998) or would you port it to OpenServer 5 (how much will
: it change in 2+ years - the prototype does _not_ run under OS5) or wait
: and see what comes out of the SCO/Unixware marriage and port to it?

I really wouldn't much care which system I'm doing development on
because I try to write my programs to be independent of the underlying
platform.  If I find things that aren't compatible when testing on a
new system, I do my best to improve my program to avoid anything
vendor specific.  This is particularly true when developing general
purpose application programs (as opposed to system administration
utilities).  Even with sysadmin stuff, I try to keep it as portable as
possible by using languages like perl and putting things that are
system dependent (like the location of user mailboxes) in files that
can be easily included by run-time programs.

Bill
--

UUCP:              camco!bill   2835 82nd Avenue S.E. S-100
FAX:           (206) 232-9186   Mercer Island, WA 98040; (206) 947-5591
http://www.celestial.com/
SPEED COSTS MONEY -- HOW FAST DO YOU WANT TO GO?