[Info-vax] Internationalization

[Bill Gunshannon]

On 1/1/19 2:55 AM, Dave Froble wrote:
> On 12/31/2018 9:03 PM, Bill Gunshannon wrote:
>> On 12/31/18 8:26 PM, Arne Vajhøj wrote:
>>> On 12/31/2018 1:04 PM, Bill Gunshannon wrote:
>>>> On 12/31/18 11:36 AM, Dave Froble wrote:
>>>>> On 12/31/2018 7:00 AM, Jan-Erik Söderholm wrote:
>>>>>> Well, the world does not stand still. We have a long list of
>>>>>> development
>>>>>> projects on our Alpha environment. Both updates to the current apps
>>>>>> but
>>>>>> also new applications.
>>>>>>
>>>>>> And I do not think that the Alpha offer from VSI was put together 
>>>>>> only
>>>>>> for us, so there must be a conciderable number of similar sites.
>>>>>
>>>>> I think that there may be more sites running old Alphas than
>>>>> anything else.  For sure there were enough for VSI to decide they
>>>>> needed an Alpha release of their own.  As "their own" HP is out of
>>>>> the picture.
>>>>>
>>>>> I'm rather sure that the first build of "their own" would be the
>>>>> most labor intensive.  While additional releases will have some cost
>>>>> to VSI, most likely not near as much, and so perhaps there will be
>>>>> further Alpha releases.
>>>>
>>>> Software availability is only one term of the equation.  The one
>>>> easily fixed.  The hardware is not going to be around for much
>>>> longer.
>>>
>>> Physical hardware becomes old and fragile.
>>>
>>> But emulators does not age the same way.
>>>
>>> So "Alpha's" may be around for decades to come.
>>>
>>> Similar to the market for VAX emulators.
>>>
>>>>> If there is a significant number of sites using Alpha VMS, who are
>>>>> willing to pay for support, that's chunk of change that VSI needs to
>>>>> consider and go after.
>>>>>
>>>>> You mentioned perhaps never going off Alpha.  Can you suggest
>>>>> reasons for that?  DEC made some very reliable stuff, but, in time
>>>>> the costs and efforts will increase.  I guess my question is, do
>>>>> reasons for not using x86 when it's available exist?
>>>>
>>>> A better question would be do (valid) reasons for staying
>>>> on Alpha once a viable alternative is there exist.
>>>
>>> The typical reasons for staying on old platform are:
>>> 1) some software dependency is not available on the new platform
>>
>> A valid concern.  But one that needs to be addressed before it
>> becomes a potential disaster.
> 
> Software is in some ways easier than hardware.  Most likely there is no 
> option of new VAX, Alpha, and soon itanic CPUs.  Not saying it cannot 
> happen, but the cost will be rather high, and there will be alternatives.
> 
> But best thing is, software does not break.  What it did yesterday, it 
> will do today, and tomorrow.

The software will not do tomorrow what it did today if the hardware
it runs on is dead.  Emulators are cool, but a very inefficient system.
One who knows their hardware is going away would  be better served
by starting the migration away from that hardware before they jump
on the emulator bandwagon.

> 
>>> 2) some special hardware is not available on the new platform
>>
>> Wouldn't that preclude the use of an emulator?  But, same as
>> above.
> 
> Not necessarily, but this is not an area I'm familiar with.

Where do you plug your custom UNIBUS device into on SIMH?

> 
>>> 3) migration cost too high
>>
>> Now high does it have to be to risk being shot out of the water
>> by an unrepairable system failure?
> 
> Methods to avoid this can be devised.

How?  As I have stated, I have stored equipment with the plan of
using it in the future only to find it died on the shelf.  Don't
know why, but I learned  not to trust this for critical systems.

> 
>>> #1 will come in effect if some vendors supporting VMS Alpha
>>> will not support VMS x86-64. I believe both PL/I and Ada
>>> may end up in that category.
> 
> But, it's not something that just cannot be done.  There is the issue of 
> cost.
>

Continuing to support obsolete hardware is a money pit.  The ROI
decreases rap[idly.  Especially if people start doing what you
hoped they would do in the first place and start migrating to
X86-64.  As the numbers of holdouts decreases so does the money
coming in to pay for that continued work.


>> Then it becomes time to look at moving off VMS.  Or changing
>> language before it becomes a critical problem.
> 
> I really, really like the way some people throw this out.  Sometimes 
> it's not feasible.  Like in "the operation was a success, but the 
> patient died".  It's really interesting how some people are so free with 
> other people's money.  Ever hear about "biting off more than you can 
> chew"?  Ever hear about "choking on it"?
> 
> I've seen companies die because of this.  Yeah, it's not your company, 
> what do you care?
> 
> Ever hear about "being practical"?

Ever hear about a catastrophic failure with no warning.  Or rather
with a warning that you chose to ignore.


> 
>>> #2 will probably be rare. I don't think there were ever as much
>>> custom hardware for Alpha as for VAX. And in case of #2 then
>>> emulators are not an option either.
> 
> Lots of stuff available for PCs.

And lots of companies that can make custom hardware, probably
at a much smaller price than for a VAX.

> 
>> Guess it depends on the Alpha.  The Alpha I had (a gift to the
>> school from Compaq) that ran VMS had a PCI bus so any hardware
>> that fit in there would fit in a x86 box as well.  Then it
>> is just a software problem.  But, as you said, not as likely
>> with the Alpha as it was with the VAX.
>>
>>>
>>> #3 will depend on VSI's approach. Just rebuild and it will run
>>> then #3 is not an issue. If breaking too much then it is
>>> another story.
>>
>> If you can't migrate to X86-64 VMS it is probably time to
>> evaluate your whole operation.  Staying in the 20th century
>> is unlikely to be a wise decision.  Unless all the important
>> people plan to retire soon and don't really care about the
>> long term future of the company.
> 
> Sigh, please read above ....

What's to read.  The writing on the wall?  Cost to do something may
appear to be expensive.  But waiting could be catastrophic.  And the
cost becomes more manageable if you start now and spread it out over
a period of time instead of waiting til you need it done today.  But
it appears no one bothers with risk management any more, so what ya
gonna do.

> 
>>> My guess is that the vast majority of remaining
>>> Alpha users will migrate.
>>>
>>> I just suspect that the migration will be very slow.
>>>
>>> Those that migrate fast already have migrated to I64.
> 
> This I think is key.  The reasons will most likely be unique for some. 
> No solution until the problem is well understood.
> 
>> Possibly, but I suspect many saw the I64 as a dead end
>> from the start and may have opted to not waste the money
>> migrating knowing if they did another migration would be
>> in their not too distant future. Either to the replacement
>> for I64 which we now see, or away from VMS.
> 
> For us, it was mainly compile, link, test, and run.  I'm not too sure 
> why there are those who cannot.  Perhaps bad decisions long ago.  But, 
> VMS has been rather good about this capability.  I'd guess they'd only 
> get better at it. >
>> Like it or  not, the choices are going to be rather simple.
>> Migrate to X86-64 VMS or migrate away from VMS or plan
>> for a demise in the not too distant future.
> 
> Or find a better choice.

Huh?  I covered both staying with VMS and leaving VMS.  What other
choice is there?  :-)

> 
>> As I mentioned in a previous post, it would be very interesting
>> to learn why VMS was chosen for a lot of the systems that are
>> still on it.  VMS was never the only choice.  If it was chosen
>> for a reason is that reason still valid today?
> 
> If it is the only environment that will run your applications, is that 
> good enough?

With the exception of something requiring special hardware (and even
that isn't really a job stopper) just what task can VMS do that no
other OS in the industry can do?  Note, I said task,  not something
like your particular application that in its current state can only
run on VMS BASIC.

> 
> I just wish those sounding like people from the big accounting and 
> consulting firms would "get a clue".  Most of those firms have people 
> who have never did anything, nor know anything, and are listened to by 
> the clueless.  The people who's parents built something, and because 
> they get put in charge, feel they actually know anything.  Many times 
> they don't, they listen to the clueless, and they lose.


bill