[Info-vax] inertia or fundamentals about langages?

[Dave Froble]

On 5/21/2019 11:13 AM, Jan-Erik Söderholm wrote:
> Den 2019-05-21 kl. 16:59, skrev Dave Froble:
>> On 5/21/2019 6:35 AM, gérard Calliet wrote:
>>> I extract a part of the usual discussion about porting or not porting,
>>> with the usual suspects.
>>>
>>> Because there I understand clearly why I don't agree.
>>>
>>>  >> If a business is running on VMS and a major part of that business is
>>>  >> a web server front end and they have to move that front end to a >>
>>> different platform where is the incentive to leave anything still on >>
>>> the VMS system?
>>>
>>>  >>>> The cost and the effort getting the apps ported off of OpenVMS,
>>>  >>>> usually.  Inertia.  This is the VSI market for the foreseeable
>>>  >>>> future.
>>>
>>> A port is a translation. Thinking any translation has'nt any other cost
>>> than (time, effort,...) goes with the prejudice that all langages are
>>> totally equivalent, and so that total translation is always possible.
>>>
>>> This prejudice is largely considered as wrong in linguistics, and,
>>> transferred in computer sciences - if you say "computers are another
>>> domain" - it could work theoritically only with pure formal langages
>>> (and again if we forget the fundamentals about termination problem).
>>>
>>> A software is always some sort of langage used to cope with some
>>> specific issue. The more the software has history, the more the its
>>> "langage" is idiomatic, and the more the total translation is just a
>>> dream.
>>>
>>> The dream of a total translation is however always present. It is one if
>>> the reason each decade brings a new "totally universal" solution, with
>>> which every tool, object, action, ressource could be unified. In a sense
>>> Cloud is our "unique" realm these years, as virtualisation has been 5
>>> years ago, SOA before...
>>>
>>> One concrete consequence of this dream is that there are on any site a
>>> lot of concurrent "unique" realms.
>>>
>>> And I agree that any company which gets now VMS is more concerned by the
>>> difficulty of integration of VMS in its
>>> not-so-unified-but-will-be-the-future site than by its intrinsik
>>> qualities.
>>>
>>> Inertia is not at all the major reason not porting out of VMS. The major
>>> reason is that translating a well adapted langage to its situation to a
>>> x general langage is just an error, based on an oversight of some
>>> fundamentals.
>>>
>>> My opinion is the opposite about inertia. The dream of an universal
>>> translation is a lack of effort thinking about the necessity of trading
>>> with different things different ways.
>>>
>>> There are perhaps 2 symetric dreams: (1) thinking My langage Is The
>>> Langage [example: some VMS[unix,as400,pascal...] fanatics are like that
>>> :) ] (2) thinking Every Langage is translatable on Every Langage, the
>>> Universe should be Unified [Example: Every thing will be in the Cloud].
>>>
>>> The VMS future is not at all because for some time inertia will protect
>>> it from other choices. It is because some things will remain on it
>>> because VMS is more adapted to their problem, and because the "langage"
>>> which was invented on it for x purpose is the better now for that
>>> purpose.
>>>
>>> 2 types of work are involved in the future:
>>> (1) understanding the parts of the software that need VMS, and the parts
>>> which would be better worked elsewhere, which is to say understanding
>>> how the specifics of the use match the specifics of VMS and central
>>> parts developed on it,
>>> (2) being able to "translate" the parts of the software that are
>>> translatable, which is creating the "interfaces".
>>>
>>> All this said, it is perhaps why "some like it hot" and some not. On my
>>> side I like it (VMS) :)
>>>
>>> Gérard Calliet
>>>
>>
>> A "port" is, by definition, a sideways move.  At best it is equal.
>> Usually there are losses, and the port is actually a backward move.
>>
>
> That is of course not true. In some cases a port might go wrong (just
> as any project) and you end up with something worse. But in most cases
> you port to something that is faster, cheaper and easier to go forward
> with. Even if the port as such does not add any functionality. The real
> question is if those benefits actually justifies the costs of the port.
>
>

Ah, Jan-Erik, we begin another fine discussion.

As I mentioned, the "port" part of any port is just a sideways move. 
Perhaps additional functionality may be added, but that is strictly 
speaking not part of the "port".

Frankly, I've never seen any port that was the equal of the old system. 
  That is just very hard to do, usually because of the lack of similar 
functionality.  Another big issue is the capability of the "porters" 
being able to understand everything about the original system.  Very 
often things are missing in the new system.

Faster, cheaper, easier.  Perhaps.  And perhaps not.

But you are correct in "benefits actually justifies the costs of the 
port".  Rarely is this true.  Ports are not easy, and not cheap.

-- 
David Froble                       Tel: 724-529-0450
Dave Froble Enterprises, Inc.      E-Mail: davef at tsoft-inc.com
DFE Ultralights, Inc.
170 Grimplin Road
Vanderbilt, PA  15486