[Info-vax] Modular VMS, was: Re: Prism/Pillar, was: Re: inertia or fundamentals about langages?

[Stephen Hoffman]

On 2019-05-29 18:11:25 +0000, Simon Clubley said:

> Nice as it would be, I'm not looking to replace VMS with a microkernel.

L3 and L4 were among the first designs that got messaging tolerably fast.

But no, VSI isn't doing research and won't be for some years, and has 
more than enough other work to do.

> I just want VMS to have the same level of module based flexibility that 
> other current monolithic kernel designs have.

The speed of the reboots—though the default Itanium self-test is 
glacial, by current standards—helps reduce the pain of the code reloads.

That and the System Code Debugger helped reduce testing- and 
debugging-related overhead.  This whole area could be improved further, 
certainly.

There are also fewer folks doing OpenVMS driver and exec work and that 
for various reasons, and the preferred long-term would be to reduce the 
requirements for that work yet further.

Adding FUSE support is one of the more interesting potential 
enhancements, though.

Semi-related, here's somebody looking at a user interface:
  https://medium.com/@jasonyuan/introducing-mercury-os-f4de45a04289

And semi-related, OpenBSD folks looking at adding unveil() to the 
existing pledge() app security support:
  https://www.openbsd.org/papers/bsdcan2019-unveil/mgp00001.html


-- 
Pure Personal Opinion | HoffmanLabs LLC