[Info-vax] Developing VMS applications on VMS or on Linux/Windows ?, was: Re: x86 Cross Tools Kit

[Stephen Hoffman]

On 2019-03-03 21:57:27 +0000, already5chosen at yahoo.com said:

> On Wednesday, February 27, 2019 at 8:35:18 PM UTC+2, Bill Gunshannon wrote:
>> On 2/27/19 9:49 AM, John Reagan wrote:
>>> On Wednesday, February 27, 2019 at 8:53:12 AM UTC-5, Simon Clubley wrote:
>>>> Should VMS development continue on VMS, with its more limited 
>>>> development environment or should VMS be treated more as an embedded 
>>>> system and development done under the richer development environment on 
>>>> Linux/Windows ?

That's the inevitable fallout from the server-only strategy, if 
developers want GUI-based development tools.  OpenVMS developers have 
been command-line, but that approach is fading among developers.  And 
having used some of what's available, that command-line approach is 
fading with very good reason. Some of the current-generation IDEs are 
vastly better than the OpenVMS command-line tools.  And VSI has a pile 
of work underway—and a far larger pile pending—to upgrade OpenVMS as a 
server, too.

>>>> You can argue this either way. What do the people here think ?

read: status quo development processes and tools, or GUI development 
tools and processes.

Many in the EDT compile-link-run-debug crowd won't be prised off EDT in 
the near future.  If ever.

If EDT works for you, go for it.

But then learning new tools and new APIs and new approaches is also 
part of working in this business.

Command-line development isn't a growing market for developers, nor for ISVs.

And  having used IDEs in recent times, I find command-line development 
slower.  Much slower.  Part of that is better frameworks and tools, but 
just code development and code-refactoring is slower.  Much slower.

>>>> Oh, and before someone says that means you need a second system for 
>>>> development, remember you are likely already using a second system to 
>>>> run your terminal emulator.

Been the case for years, too.  The DEC VT terminal business got 
flattened in the 1990s, obviously.

>>> Well, Linus says that you can only develop for x86-servers from x86 
>>> platforms and that moving/porting/changing platforms is bad.  (I guess 
>>> Apple was lucky? for their ports from MC68040->PPC->x86->ARM)

I read the original posting rather differently.

https://www.realworldtech.com/forum/?threadid=183440&curpostid=183486

What Linus is referencing with Linux and x86-64 is relevant to the 
situation that OpenVMS has found itself in, in recent years.  Not 
having any entry-level; hardware or software.  Having weird and 
expensive hardware.  Unusual APIs and outdated and unstable development 
tools.  Not having much in the way of hobbyists, either.  As for 
porting, some customers and some apps have fallen off of each OpenVMS 
port.  That all sounds familiar, right?

Having the hardware and the tools with the hobbyists and with the 
smaller organizations got both Windows and Linux where they are now.

And for now, that's also often on x86-64 hardware.

Whether the hobbyists stay with that?  Arm has massive production 
volume already, and which is unusual for any potential x86-64 
competitors in the past couple of decades.

Will we get Arm servers?  We already have.  We have zillions of them.   
Little tiny ones.  Arm boxes and more than a few tiny Arm-based servers 
folks are working with, with RPi and equivalent.

Will those tiny little servers and those embedded computers displace 
the bigger rack-mount x86-64 servers for the folks running Windows 
Server or RHL?  Of course not.  Not for many years.  If ever.

But those little tiny configurations will pick up new markets and new 
apps.  They already have.

Just as mobile phones have gotten to billions of devices with Arm, and 
have created whole new apps and new capabilities and new markets.

With most of that mobile market running Linux or Unix, too.

Will Windows and Linux be on Arm boxes? Both already are.

Will those little tiny Arm boxes get bigger and faster?  Absolutely.

That progression from little tiny boxes to larger and more capable 
boxes and to wider acceptance is what happened with Windows and with 
Linux, too.

> ....what's Linus really said is not stupid, even if I personally disagree.

Well, that's reasonable.  Linus was disagreeing with you, too.

> He is strongly in favor of the same or very similar platform for 
> development and deployment, as most people in that posted in this 
> thread.

I'd like that, too.

But there isn't a competitive GUI on OpenVMS, there are no front-end 
tools, and there isn't (yet) common hardware, and there isn't (yet?) 
current software.

And the software licenses will run you several thousand dollars at 
present, and with the x86-64 pricing TBA.

Among other limitations.

With no recent GUI front-end and no recent tooling and with few 
(documented) hooks for third-party development tools, native IDEs are a 
tough project.

And Java on OpenVMS isn't all that speedy for any code that doesn't get 
JIT'd, for the folks working with applications based on that tooling.

With x86-64, you're at least on the same hardware as your server, and 
can be on co-resident on the same server box with a hypervisor.  But 
that's all still working cross-platform.

> Even El reg article itself is not too stupid despite author's partial 
> misrepresentation and misunderstanding of Linus's postion.
> What is really stupid is a headline of El Reg article.

So go to the original source.  In this case, that'd be to Linus' 
response to your comment.  Get some context.

OpenVMS?  It's all about VSI, and all about the installed base.  Beyond 
the retro-computing and computing-nostalgia folks, and beyond the 
installed base, few ISVs and few hobbyists are soon interested in 
OpenVMS.  Not pending a whole lot of work at VSI, and with the earliest 
parts of that underway.  And VSI doesn't have a competitive entry-level 
hardware and software offering, though the x86-64 support will help 
with that.  There's that "fun" balancing act here between (current) EDT 
folks and (future) IDE folks, too.

TL;DR: both Windows and Linux are of the scale to aim at 
mobile-to-cloud, and with hobbyist- and ISV-friendly hardware and 
software.  Microsoft keeps trying on mobile, and they're having "fun" 
with Amazon at the other end.  OpenVMS has no shot at that whole range, 
and VSI is left to pick what parts and products they can build and 
deliver.




-- 
Pure Personal Opinion | HoffmanLabs LLC