[Simh] A new version...

[Terry Newton]

--- On Fri, 9/19/08, Philipp Hachtmann <hachti at hachti.de> wrote:

>  > I was surprised that the new
> > make is much faster than using the stock makefile.
> 
> Didn't I tell?

Yes.. it's true.. but speed of compile matters only if
developing for simh, and only if recompiling the whole
thing. So for these folks it's a good thing! I tend to
only compile the whole thing once when a new sim comes
out, maybe tweak a few of the sims I use regularly by
recompiling with different compilers etc and integrate
my terminal patches into the HP2100 sim. Speed is *not*
an issue for me, when working on just HP2100 it's the
difference of 9 seconds (timed it) vs a couple seconds.
It's not like I do it all the time. With the latest
version of my terminal mod I added switches to enable
and disable the translations, recompiled maybe 10-20
times getting it right - over many hours of coding
and study the compile time amounted to maybe 3 minutes.

What is important to me, is after putting in my patches
I have to compile it under Windows to generate hp2100.exe
to include on my hp2100mod page, and it would really
aggrevate me if I got to that step and it no longer
worked, so that's why I felt compelled to speak up
about the wisdom of leaving the default makefile alone.
Because it works. Not everyone has a fancy computer,
and frankly I see far more troubles reported by those
with the latest and greatest than with modest hardware.
Soon I need to get a new machine, and I dread it.

And no, not giving virtual XP any more ram... it shall
deal with 192 megs or crumble. MinGW/cygwin/perl etc runs
just fine in my virtual XP, running out of memory has not
been an issue with any other apps I run there. In fact
this was the first time I've seen such an error message.

> > The only
> > thing I need to do that I didn't have to do before
> is manually
> > remove the object files... (very carefully) rm *.o
> followed by
> > rm */*.o - this is acceptable since it gives the
> choice to
> > leave them and speed up recompiles.. 
> 
> Why do you do that manually? "make clean" does
> that for you.

That also removes the executable binaries. The BIN dir
*is* the install and I don't want to remove it. On my
system I keep a "simh" shortcut pointed to the latest
simh version, all my shell scripts for sims are coded
to run [partitiondir]/simh/BIN/simname. I could simply
rename BIN temporarily then run make clean but if I use
the new makefile a lot I can simply put the .o remove
commands in a shell script and be done with it. This only
needs doing at the end of a hack session so whatever.

> Have fun with HP21MX! Eh, what are you doing 
> for the HP? I have two E-Series machines without any idea
> what to do with them...

There's HP-IPL/OS, an operating system Bob Shannon and I
wrote, has support for 7900 and 7906 disk drives and if
into soldering can use an older-style IDE disk. HP-IPL/OS
can run most ABS (absolute binary system) files from disk,
including things produced using BCS and HP/MSU-BASIC.
At least 64KW of memory is required, 32KW for the app
and 32KW to hold HP-IPL/OS while the app is running or
to buffer the "alien" binary. On a smaller scale you
could rig up a PTR emulator and at least load code.
This page might give you some ideas...

  http://newton.freehostia.com/hp/

Recently Bob turned me on to the Vinculum series of USB
interfaces with built-in support for a FAT file system,
permitting using a thumbdrive or USB hard drive with
a suitable interface adapter. In the planning stages,
nothing definite yet but hopefully will support a 7906
drive emulation using a 13037 controller interface.
Another potential app for the USB chip would be a
papertape reader/punch emulator that permitted
attaching to named files, in effect a crude dos.

But at the moment my dream HP app is a Z-machine
interpreter for running Zorks, HitchHiker's Guide,
Planetfall etc. Seeing these run on a E-series mini
would be inspiring...

Terry