OpenQM and jBase

[Donald Montaine]

2 `1/2 months after the announcement and both the QM and jBase websites still refer to Zumasys, no reference to Rocket.  I understand where jBase fits into the mix   But can not figure out where QM fits.  It's main purpose was to allow users to maintain their software on a platform with a lower yearly cost than D3 or U2.  But why would Rocket want to market QM at all now?  

[Steve Trimble]

That is my fear as well Donald.

I really love QM. Its features and support have been like no other of the multi value players. At least, to me.

Here's keeping my fingers crossed.

Computerized Data Mgmt Inc
Steve Trimble
(501) 772-3450 cell / text

cdm...@gmail.com

On Wed, Jan 5, 2022 at 5:24 PM Donald Montaine <dmon...@gmail.com> wrote:

2 `1/2 months after the announcement and both the QM and jBase websites still refer to Zumasys, no reference to Rocket.  I understand where jBase fits into the mix   But can not figure out where QM fits.  It's main purpose was to allow users to maintain their software on a platform with a lower yearly cost than D3 or U2.  But why would Rocket want to market QM at all now?  

--
You received this message because you are subscribed to
the "Pick and MultiValue Databases" group.
To post, email to: mvd...@googlegroups.com
To unsubscribe, email to: mvdbms+un...@googlegroups.com
For more options, visit http://groups.google.com/group/mvdbms
---
You received this message because you are subscribed to the Google Groups "Pick and MultiValue Databases" group.
To unsubscribe from this group and stop receiving emails from it, send an email to mvdbms+un...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/mvdbms/25bf7a9c-d8ea-4169-bf90-cb454361e02an%40googlegroups.com.

[Donald Montaine]

I agree.  I love QM but have largely moved on from multi-value altogether..  The yearly licensing model just does not fit what I want to do.  When it was 10 yrs of support with Accuterm for a small extra fee at the time of purchase, it was a perfect fit for me.  Now, with yearly support and Accuterm licensed separately, it is no longer a good fit for my purposes.  The pricing and licensing orientation is the same as for all the other currently maintained MV products and, for me, is no longer competitive with other options.  Unfortunately, the attempt to move ScarletDME to 64 bit has never progressed, or that might have been a possible alternative.

[geneb]

On Wed, 5 Jan 2022, Donald Montaine wrote:

> competitive with other options. Unfortunately, the attempt to move
> ScarletDME to 64 bit has never progressed, or that might have been a
> possible alternative.
>

All we need is one C guru to make it happen. ;)

g.

--
Proud owner of F-15C 80-0007
http://www.f15sim.com - The only one of its kind.
http://www.diy-cockpits.org/coll - Go Collimated or Go Home.
Some people collect things for a hobby. Geeks collect hobbies.

ScarletDME - The red hot Data Management Environment
A Multi-Value database for the masses, not the classes.
http://scarlet.deltasoft.com - Get it _today_!

[Donald Montaine]

Well, if enough people actually cared, we could do the same thing other open source projects do, offer a bounty to attract a C programmer.  But the sad thing is that the MV community doesn't seem to be motivated to do anything to save the technology from the "let's milk it 'till it dies" crowd.

[geneb]

Agreed. :(

[Will Johnson]

I've always thought that Rocket was much too hands off.  But that's a pattern in MV in general.

Most places I've worked, they had no idea, and didn't care about the O/S at all.

They would post ads like "Looking for a Pic Programmer" (sic) in a local newspaper, because they had no idea what it was or how to go about finding a new programmer

All communications, are through a VAR or independent consultant.

Each place, I try to reconnect the people to Pick, Adds, Rocket, VMark.. or whomever.

I just fear that each time I leave, they revert to knowing nothing.

So that feeds the idea that it's "old" and needs to "go"

That's the embedded problem.

Rocket has outreach in a sense.  You have to *find* them, then they reach out.

The MV Vendors needs to tunnel through the vars and communicate directly to the end users

[Jim Idle]

I think you would need to move the source code to github, probably change the license to Apache, update the web site and so on. That project looks abandoned as it stands and I would not bother to install subversion just so that I could take a look at the source code.

--
You received this message because you are subscribed to
the "Pick and MultiValue Databases" group.
To post, email to: mvd...@googlegroups.com
To unsubscribe, email to: mvdbms+un...@googlegroups.com
For more options, visit http://groups.google.com/group/mvdbms
---
You received this message because you are subscribed to the Google Groups "Pick and MultiValue Databases" group.
To unsubscribe from this group and stop receiving emails from it, send an email to mvdbms+un...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/mvdbms/alpine.LRH.2.21.2201060609200.1181%40sidewinder.deltasoft.com.

[Donald Montaine]

The source code is already on github.  A fork that has some modifications toward 64 bit is at https://github.com/geneb/ScarletDME.  The unmodified version is at https://github.com/FOSS4MV/ScarletDME.

Header from original source code.

* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation; either version 2, or (at your option)
* any later version.

Since it was released by the original author as GPL, any modifications would have to remain under the GPL.

License could not be changed without the original author's permission.

[Donald Montaine]

I set up a Debian 11 32 bit (Crunchbangplusplus) virtual machine.  The unmodified code from 9 yrs ago does not compile.   However the ScarletDME code by geneb does compile and I was able to get a ScarletDME system working without problems. I used the setup document, "Installing OpenQM (GPL) on Ubuntu Linux", by Rush Flat Consulting  for a very old version of Ubuntu as my setup guide, making corrections to the process when necessary. I did find a couple of minor issues during setup and have a couple of suggestions.  I am planning to write a bash script to automate much of the setup.  Note: geneb, I sent you a contact request on LinkedIn. 

[Steve Trimble]

sweet Donald!

Computerized Data Mgmt Inc
Steve Trimble
(501) 772-3450 cell / text

cdm...@gmail.com

To view this discussion on the web visit https://groups.google.com/d/msgid/mvdbms/959f7724-58fe-4494-b44c-d5b5bdd38ca1n%40googlegroups.com.

[geneb]

On Fri, 7 Jan 2022, Jim Idle wrote:

> I think you would need to move the source code to github, probably change
> the license to Apache, update the web site and so on. That project looks
> abandoned as it stands and I would not bother to install subversion just so
> that I could take a look at the source code.
>

My apologies Jim, I'd completely forgotten to update the wiki since I
moved the source code to github a few years ago. The ScarletDME source
code has been hosted on github for years. Here's the project url:
https://github.com/geneb/ScarletDME

I'll try to go over the wiki and get it updated so folks are pointed to
the git repo and not svn.

The license cannot be changed without Martin's ok, which I would
absolutely love to get - are you up for that Martin?

tnx.

[TOM MARRACCI]

Gene,

Just out of curiosity, how did you get around the 32 bit p-code issue?

This was a big stumbling block for me porting ScarletDME to 64 bit mac some time ago and without source I could not come up with a workaround.

Tom

> --
> You received this message because you are subscribed to
> the "Pick and MultiValue Databases" group.
> To post, email to: mvd...@googlegroups.com
> To unsubscribe, email to: mvdbms+un...@googlegroups.com
> For more options, visit http://groups.google.com/group/mvdbms
> --- You received this message because you are subscribed to the Google Groups "Pick and MultiValue Databases" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to mvdbms+un...@googlegroups.com.

> To view this discussion on the web visit https://groups.google.com/d/msgid/mvdbms/alpine.LRH.2.21.2201070702050.15137%40sidewinder.deltasoft.com.

[Donald Montaine]

Re: changing the license.  OpenQM was copyrighted by Ladybridge, not Martin.  So, my guess is that any change in license would have to be approved by Rocket at this point, unless the OpenQM copyright was not sold to Zumasys.

Also, I suggest that further conversations about ScarletDME development be posted on the ScarletDME google group so as not to clutter up this group with postings that may not be of interest to many members.

If you want to contact me directly, email me on my g mail account: dmontaine.

[Marcus Rhodes]

I had the same problem, but on Linux.  Try my installation script: http://www.thinqware.com/services_training_install_openqm.html

I still use ScarletDME on my 64-bit Linux boxes every day, and this is how I installed it.  It worked last time I tried it, but, of course, ymmv. Take notes and let me know what changes I need to make, if any.

[TOM MARRACCI]

I joined the ScarletDME group. Do my eyes deceive me? Is the last post from 2018?

Tom

To view this discussion on the web visit https://groups.google.com/d/msgid/mvdbms/9aee7c04-4ab6-4922-80d2-508c65f4cd31n%40googlegroups.com.

[TOM MARRACCI]

Marcus,

You can run 32 bit apps on a 64 bit machine (except for the latest Mac if I understand correctly)

What you cannot do is compile a new version with 64 bit as the calling mechanism is completely different.  32bit Intel uses a stack to make calls, whereas the 64 bit Intel uses registers for the first 7 or 8 arguments (I cannot recall how many exactly)

That plus the registers used for calling are 64 bits wide and when any calls are made to the precompiled p-code, you get a crash.

Tom

To view this discussion on the web visit https://groups.google.com/d/msgid/mvdbms/2ca20ad1-9482-4a8a-b73e-721b338a12b8n%40googlegroups.com.

[TOM MARRACCI]

I would like to keep the conversation going on ScarletDME group, but I don’t have permissions to post, so I will continue the conversation here.

I reviewed my notes and found that I was able to compile the system to 64 bit with myriad changes that I cannot recall (it’s been a while).  I can get the server to start, but when I try to connect a session, I get an error "Pcode item AK not found”

Not sure what that means or what to do next. I tried to disassemble p-code in an effort to reassemble it in 64 bit, but found that I could not.  Not sure what object language it actually is, but at first glance it doesn’t look like intel machine code.

Tom

To view this discussion on the web visit https://groups.google.com/d/msgid/mvdbms/56D6E7C5-8BA2-4625-AB40-E218B82FBE07%40aircraftspruce.com.

[Wols Lists]

Well, I care, and I'm a bit of a C programmer ... what I can do though,
I don't know. Having seen the other post about calling conventions, etc,
we should be able to force it somehow, just need to work out how ...

Cheers,
Wol

[Wols Lists]

On 07/01/2022 17:28, 'TOM MARRACCI' via Pick and MultiValue Databases wrote:
> I reviewed my notes and found that I was able to compile the system to
> 64 bit with myriad changes that I cannot recall (it’s been a while).  I
> can get the server to start, but when I try to connect a session, I get
> an error "Pcode item AK not found”
>
> Not sure what that means or what to do next. I tried to disassemble
> p-code in an effort to reassemble it in 64 bit, but found that I could
> not.  Not sure what object language it actually is, but at first glance
> it doesn’t look like intel machine code.

No. P-code is short for pseudo-code. So it's a form of machine code, but
it runs on a "virtual cpu" which is a program running on a real cpu.

That's how things like Pascal used to run on a big variety of systems,
you wrote a virtual cpu that ran on whatever hardware you had, and that
ran any code you got from anywhere else.

What p-code this is, I don't know, but if Martin had access to Pr1me
documentation, it could be the variant used there. Compile INFORMATION
code would run on any INFORMATION system, and that included (and may
have been limited to) 50-series Pr1mes, MIPS-powered EXL systems, and
PC-INFO which ran on a Xenix 386. (And I believe the 50-series cpus had
a microcode mod that let them run p-code directly).

But anyways, this means that whatever is running the p-code is actually
an interpreter. Fix the variable-size to whatever it is expecting, and
it should work fine.

Cheers,
Wol

(I'll need to try and get this all running at some point :-)

[TOM MARRACCI]

Looks like a slight ambiguity with the p-code file itself.

The version that I have from ScarletDME is 29748 bytes long. The one that I have from OpenQM when I demoed is 46752 bytes long. When I stepped the program with gdb, only 2 modules were inspected looking for the AK module which I’m pretty sure would not be correct.

Tom

> --
> You received this message because you are subscribed to
> the "Pick and MultiValue Databases" group.
> To post, email to: mvd...@googlegroups.com
> To unsubscribe, email to: mvdbms+un...@googlegroups.com
> For more options, visit http://groups.google.com/group/mvdbms
> --- You received this message because you are subscribed to the Google Groups "Pick and MultiValue Databases" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to mvdbms+un...@googlegroups.com.

> To view this discussion on the web visit https://groups.google.com/d/msgid/mvdbms/ade3f612-f8eb-3003-a8cb-f5a5cd068b0c%40youngman.org.uk.

[TOM MARRACCI]

Thanks for the explanation of the p-code. That makes perfect sense. It seems the Pcode function in qm.c needs some work here to interpret the file correctly.

Tom

> --
> You received this message because you are subscribed to
> the "Pick and MultiValue Databases" group.
> To post, email to: mvd...@googlegroups.com
> To unsubscribe, email to: mvdbms+un...@googlegroups.com
> For more options, visit http://groups.google.com/group/mvdbms
> --- You received this message because you are subscribed to the Google Groups "Pick and MultiValue Databases" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to mvdbms+un...@googlegroups.com.

> To view this discussion on the web visit https://groups.google.com/d/msgid/mvdbms/9846ea50-7908-ee69-d0b9-4aa5dc6cb00a%40youngman.org.uk.

[Donald Montaine]

The ScarletDME group.  I asked to join this morning and now have write access.  So anyone wanting to join should be able to write once you are approved.

[Wols Lists]

On 07/01/2022 17:40, 'TOM MARRACCI' via Pick and MultiValue Databases wrote:
> Thanks for the explanation of the p-code. That makes perfect sense. It seems the Pcode function in qm.c needs some work here to interpret the file correctly.

I'm guessing (and it seems probable) that the structures used to import
p-code into the interpreter contain a load of fields defined as "int".
NOT wise. They need to be found, and converted to int16 or int32 or
whatever is appropriate.

Cheers,
Wol

[geneb]

On Fri, 7 Jan 2022, 'TOM MARRACCI' via Pick and MultiValue Databases wrote:

> Gene,
>
> Just out of curiosity, how did you get around the 32 bit p-code issue?
>

It's an issue that solved itself once the data types were fixed. If you
check out the "dev" branch at the git repo, you can build it for 64 bit
and it kind of works. I suspect there's some pointer magic going on that
is making it unstable. You can see what the current issues are in the
change_log.txt file. (entries are dated 28/29 Feb 20).