[Info-vax] New CEO of VMS Software

[Arne Vajhøj]

On 1/4/2024 8:48 PM, Lawrence D'Oliveiro wrote:
> On Thu, 4 Jan 2024 20:26:33 -0500, Arne Vajhøj wrote:
> 
>> On 1/4/2024 5:20 PM, Lawrence D'Oliveiro wrote:
>>> On Thu, 4 Jan 2024 15:42:57 -0500, Arne Vajhøj wrote:
>>>> On 1/4/2024 2:25 PM, Lawrence D'Oliveiro wrote:
>>>>>
>>>>> ... put in place the program I suggested sometime back: get rid of
>>>>> most of VMS itself, leaving only the parts that users care
>>>>> about--namely their userland programs and DCL command procedures. All
>>>>> that could run on an emulation layer on Linux.
>>>>
>>>> Lots of work to implement.
>>>
>>> Much less than the 7 years it took to reimplement VMS on top of AMD64.
>>
>> I doubt that.
>>
>> Mapping from one OS to another OS is not easy.
> 
> Linux is a more versatile kernel than VMS. For example, the WINE project
> has been able to substantially implement the Windows APIs on top of Linux,
> while Microsoft’s attempt to do the reverse, implement the Linux APIs on
> top of the Windows kernel with WSL1, has been abandoned as a failure.

Excellent examples.

Have you noticed how the world has moved from Windows to Linux
with Wine? No. Because it did not happen. Wine is a niche
thing.

MS tried WSL1 and changed to to a VM model with WSL2.

2 x commercial failure.

>>> Remember, it took less time (and resources) than that to move Linux
>>> from 32-bit x86 to 64-bit Alpha.
>>
>> Very different task.
> 
> How different? It’s exactly the same sort of thing: port an OS to a new
> architecture.

If you call both a CPU and an underlying foreign OS kernel for "a new
architecture" then yes.

But the reality is that it is very different.

>>>> the VMS customer base.
>>>
>>> Maybe they have.
>>
>> That is something we would know about.
> 
> You mean “would not know about”?

No. We would know.

A company could not pick a large number of DEC/CPQ/HP/HPE/VSI
customers without the VMS community knowing.

>> They have customers, but not nearly as many as those migrating
>> natively to other platforms.
> 
> I think we’ve discussed their product before. Reading between the lines of
> their case studies, seems their product lacks some of the niceties that it
> should be possible to implement on top of the Linux kernel. DECnet, I
> think, was one thing they seemed to be missing.

They do run on Linux (and Windows).

It is possible that someone could do better than them.

But they did not.

And there were a couple of other companies offering
similar (or somewhat similar) services: Accel8 and BosBC. They
are no longer in business.

That makes it a 0 out of 3 success rate.

Arne