[Info-vax] BASIC compiler in the hobbyist distribution

[Stephen Hoffman]

On 2015-05-29 11:37:04 +0000, lists at openmailbox.org said:

> On Thu, 28 May 2015 20:04:41 -0400
> Stephen Hoffman via Info-vax <info-vax at rbnsn.com> wrote:
> 
>> On 2015-05-28 23:26:16 +0000, David Froble said:
>> 
>>> seasoned_geek wrote:
>>>> On Thursday, May 28, 2015 at 1:50:52 PM UTC-5, Bill Gunshannon wrote:
>>>> 
>>>>> That was my point.  Is Solaris even relevant in today's IT world?
>>>>> If I were VSI I would not waste time trying to determine how to
>>>>> "compete" with Solaris.  And AIX while still doing OK is really a
>>>>> very niche product and probably not really a competitor anymore.
>> 
>> Ayup.  I'd expect those AIX folks to head to Linux or to System z, if
>> they decided to port off of the current POWER systems.  Some few might
>> go to Solaris or BSD or to some other Unix.
> 
> AIX -> z is an absolute non starter.

The premise is folks that want reliability, and are already running 
IBM, and that are porting.  So... Yes, System z (hardware) is in play.  
Maybe not z/OS (software).

> First of all the costs and capabilities of z so far exceed AIX there isn't
> any discussion at all. If they needed z they would already be on it. AIX is
> competition for other commercial UNIX. That market is mighty thin these
> days.

Which was my point.  Not enough of those folks likely to move, and not 
enough of those that will move are going to move to OpenVMS.

FWIW, I don't make the "commercial UNIX" distinction.  Porting 
applications among the Unix platforms has been a hassle for the last 
thirty years I've seen of it.   Porting among the various choices has 
gotten somewhat easier, but it's still hassles.  Those AIX folks are 
not going to be looking to port to OpenVMS and not to OpenVMS without 
open-source — open source which seasoned_geek has been railing against.

> You might be able to port and write your stuff to move to another
> commercial UNIX or Linux. It won't be easy and it won't be cheap.

Which was my point.   It'll be cheaper than a port to OpenVMS.

> 
>>>> Actually worth it, more so than chasing the "free" market at this
>>>> point. The people in those particular niches, particularly AIX and
>>>> AS/400, have real business needs and have been able to justify
>>>> spending of significant cash to service those needs.
>> 
>> So you're suggesting that the AIX folks port from a Unix system to
>> OpenVMS?   The port starts out as pretty much a rewrite.
> 
> And your comment that you would "expect those AIX folks to head to Linux or
> to System z" doesn't entail a complete rewrite in the System z case? And as
> I said z is magnitudes more capable then anything else being discussed
> here and has nothing in common with AIX or any UNIX machine. It is not used
> to run UNIX-like work. If they need UNIX-like workloads or apps they run
> zLinux in an LPAR (roughly, a VM on z).

Which runs on System z hardware.

> If they needed  System z they would already be running that and not AIX.
> That they're not means a mainframe is not on their radar.

But OpenVMS is?

> 
>> One Egg One Basket designs have been fading out, thankfully.
> 
> Hard to tell from the numbers. IBM just announced healthy growth in the
> mainframe server (System z) market just today.

I was referring to application designs.    Critical applications that 
run on one big box — without provisions for some sort of failover — are 
getting rare.  Some of these designs involve cold or warm standby 
servers, and an increasing number of the designs can use multiple 
servers for performance or geographic distribution or whatever.

Healthy growth is good.  Healthy growth in an expanding market might 
not be quite as good.  Without numbers and trends — and given this is a 
commercial system — revenue trends, data points don't have all that 
much meaning.

> 
>>>> A much easier sell to tell the AIX and AS/400 crowd, hey, your boxes
>>>> were supposed to kill the VAX, it is still here. We have 24x7 up-time
>>>> measured in decades with ongoing development and you're looking at End
>>>> Of Life for your current platform, here is how you port...
> 
> I don't believe that at all (check quoting depth- this appears to be from
> seasoned_geek who btw does me a great service 95% of the time since I don't
> have to write anything after reading what he says- but this time I don't
> agree). AIX was designed to kill other commercial UNIX. And it did pretty
> well. Who's left?

Linux, the BSDs, and OS X, as the Unix boxes go.

> The AS/400 was the grandpa of the S/34 and S/36 and was just a general
> purpose office mini for commercial applications. What it did kill was just
> about every DEC mini. State Farm still has a farm of AS/400s running today.
> None of this is competition for VAX. It's a different market and a
> different workload.

My point: you're not likely to see that stuff ported to OpenVMS, which 
was what was being posited.
> 
>> The instructions are going to involve learning a wholly new platform,
>> and rewriting some or all of the not-COBOL and not-Fortran code
>> involved in the applications, and sorting out more than a few file and
>> database differences.  Which may well be a rewrite measured in decades,
>> for some folks.
> 
> I have lost sight what this is referring to but COBOL and FORTRAN are some
> of the toughest languages to port given how many vendor extensions there
> were, and all of them different. This often comes down to a rewrite too.

Yes, there are extensions to COBOL and Fortran.  Some of the ports I've 
done have been very straightforward.  Others required rework.

>>> A really bad bet would be to go after IBM, again, which is part of how
>>> we got to where we are now.
>> 
>> Ayup.  But then I'm also wondering what sort of hardware seasoned_geek
>> would be going after these IBM customers with, here.  Probably not with
>> Oracle SPARC.  Beyond Kittson, probably not Intel Itanium.  ARM isn't
>> fast enough (yet?).  POWER?  That would be going after IBM and IBM
>> users on their own home hardware turf.   That'd involve an OpenVMS
>> port, and then making a really tough marketing case.
> 
> Earlier I did suggest VMS should be ported to POWER. POWER is now open and
> real non-IBM servers are available. A premium OS like VMS needs premium
> hardware. Running on Intel crapware is only going to lower VMS to their
> standards. "You can't wrestle with pigs without getting dirty."

So you're proposing getting the AIX folks to port to OpenVMS on POWER?

Now for the wider and more general market, VSI just won't sell enough 
POWER boxes to matter.  POWER is a dead-end, per all the hardware 
architecture trends.  By all external appearances, IBM is spinning off, 
opening up, and otherwise getting out of that business.  Doing what 
they can to keep it going, but it's not headed in the right direction.  
 If the VSI goal is volume sales on reliable hardware — and 
seasoned_geek posted a pretty good testimonial to the general quality 
of some of those Lenovo x86 server choices — then that's x86 boxes.  
Getting folks to port their applications to OpenVMS running on another 
hardware platform that's not x86?  AMD effectively put the kibosh on 
that a while back, and Intel finished the job.   Not gonna fly.



-- 
Pure Personal Opinion | HoffmanLabs LLC