[Info-vax] RMS record metadata, was: Re: Re; Spiralog, RMS Journaling (was

[Simon Clubley]

On 2016-06-21, Stephen Hoffman <seaohveh at hoffmanlabs.invalid> wrote:
> On 2016-06-21 14:16:50 +0000, lawrencedo99 at gmail.com said:
>
>> Of these, none is worth keeping. Except maybe that last one.
>
> Your list missed one: if RMS is ripped out, then large chunks of 
> OpenVMS itself and more than a little third-party and end-user code 
> either gets rewritten. or an RMS analog layer gets written.  That 
> latter approach and that latter effort goes nowhere near forward 
> progress, either.  There are a lot of RMS system service calls around 
> in VSI and third-party and end-user source code, and a lot of source 
> code with in-built RMS-based assumptions.
>

But OTOH, if you wanted to build something which looked like VMS but
without maintaining strict VMS compatibility, RMS as it currently
stands is something which would be ripped out with a vengeance.

You would need something RMS like in a new VMS, but it needs to be a
procedural subroutine style API interface, not a control block interface.
There's way too much direct exposure of low level details to VMS user
applications; that kind of thing needs hiding behind an abstraction
layer which does some of the boilerplate stuff for you.

With abstracted interfaces, you are also less likely to have portability
issues when the underlying hardware architecture changes or the OS is
enhanced.

Simon.

-- 
Simon Clubley, clubley at remove_me.eisner.decus.org-Earth.UFP
Microsoft: Bringing you 1980s technology to a 21st century world