[Info-vax] DCL, was: Re: Microkernel

[VAXman- at SendSpamHere.ORG]

In article <k0lifr$6p2$1 at dont-email.me>, Stephen Hoffman <seaohveh at hoffmanlabs.invalid> writes:
>On 2012-08-17 11:47:29 +0000,   VAXman-  @SendSpamHere.ORG said:
>
>> In article <k0jvkr$r1q$1 at dont-email.me>, Stephen Hoffman 
>> <seaohveh at hoffmanlabs.invalid> writes:
>>> 
>>> Last I looked at this, there was a cha-cha to get the code remapped
>>> from P0 to P1 space, and some other arcana.
>> 
>> There's now an IAC${V,M}_P1DIRECT.
>
>That'd be handy.  IIRC, KP threads was retrofit into DCL.  The old 
>home-grown stuff didn't survive Itanium.
>
>> The fun part is getting from EXEC
>> to SUPERVISOR mode and hoping all goes well. ;)  Debugging there is an
>> interesting exercise in hair pulling.
>
>So that's why there were some bald heads in engineering.  Never made 
>that connection.  :-)

The clumps of hair on my floor may be a testament to that! :)



>I usually ended up using the ring buffers, broadcasts and (depending on 
>the code) delta or xdelta, and/or an SDA extension.
>
>The VMS inner-mode debugging infrastructure was (in comparison to what 
>I'm using on other boxes) weak, and somewhat scattershot.

KERNEL and EXEC are more readily debugged.  It's the funky mapping and
that it's SUPERVISOR making it more interesting.  Yes, there's the ways
you've described and then some.  DELTA, on Itanium, does work but it's
slow when referencing back through the listing.



>>> And the new CLI will want to be field the callbacks arriving from the
>>> sys$cli routine, no?
>> 
>> If you implement your own CLI would you necessarily need the features
>> of the DCL SYS$CLI?  Sure, you could implement your own callback but
>> would that same limiting mechanism suffice?
>
>
>IIRC, sys$cli was the path for spawning processes and for a few other 
>operations, so it wasn't really what I'd consider an optional API for a 
>CLI.
>
>
>>> I'm not particularly interested in extending DCL and enhancing the DCL
>>> buffer, myself.  I'll leave that to other folks.  To me, a different
>>> shell would be more interesting.
>> 
>> I'd toyed with the idea of taking the GNV bash and making it a proper
>> CLI but I've been busy with other projects that provide slightly more
>> income than freeware. ;)
>
>
>Or with other projects with a perceived higher long-term payback than 
>working on OpenVMS Freeware, or however the MBAs phrase that stuff.
>
>It'd be nice if somebody tossed the old DEC/shell code onto the 
>Freeware web site.  :-)

I'd think you, of all people, might know where that code has landed.

-- 
VAXman- A Bored Certified VMS Kernel Mode Hacker    VAXman(at)TMESIS(dot)ORG

Well I speak to machines with the voice of humanity.