Geneve OS development discussion

[+dhe]

@9640News do you remember having tutorials on 'shell' scripting in MDOS in 9640news?

 

Do you know if there is a way to do an alias in MDOS?

 

I'd like to alias cat>type and ls>dir - alternatively, maybe a wrap script?

 

 

[+9640News]

59 minutes ago, dhe said:

@9640News do you remember having tutorials on 'shell' scripting in MDOS in 9640news?

 

Do you know if there is a way to do an alias in MDOS?

 

I'd like to alias cat>type and ls>dir - alternatively, maybe a wrap script?

 

 

 

Do you mean something like:

 

DIR `P > TEST

 

As far as "shell", I am thinking you may be remembering GenBench Shell.  I don't think it did scripting at least not what you are wanting.

 

If the above example does not get what you need, sounds like a custom utility you might want to write ?

[+dhe]

I think I found what I was looking for.

On the CYC CD there is a catalog of the 9640news CDROM.

 

It's a bit scattered, but I found a number of ARK's with example MDOS "batch" programs - I tend to use the terms shell scripts / batch files / python programs / powershell / batch scripts pretty loosely and interchangeably.

 

---

 

quick test of wrapper batch file - ls

contents

dir %1

 

so if I accidentally type ls or ls e: - it works.

 

I'll make a wrapper for type <> cat and stick it in my batch directory called belfre - that's where the batch files live.

[RickyDean]

1 hour ago, dhe said:

I think I found what I was looking for.

On the CYC CD there is a catalog of the 9640news CDROM.

 

It's a bit scattered, but I found a number of ARK's with example MDOS "batch" programs - I tend to use the terms shell scripts / batch files / python programs / powershell / batch scripts pretty loosely and interchangeably.

 

---

 

quick test of wrapper batch file - ls

contents

dir %1

 

so if I accidentally type ls or ls e: - it works.

 

I'll make a wrapper for type <> cat and stick it in my batch directory called belfre - that's where the batch files live.

Bat's in the Belfre- I like it!!?

[+dhe]

An EXEC -utive Adventure.

 

 

  Long time since playing with assembly language - like 20+ years, and I never did much anyway.

 

  I'm playing some more with my wrapper script for Jeff White's HWID.

 

  It kept locking up, when executing the program image with exec, but it worked when gpl was loaded and the object code was executed with e/a.

 

   I had a feeling I might have had a fence post error in my screen clear routine, but rather than debug that, I just pulled a routine by Bruce Harrison.

 

    Now the program will run with exec or loaded via gpl.

 

    I went on to the VDPID routine. Interestingly enough - the program image when executed with exec shows the system as having a 9918A.

 

    Execution via GPL+E/A of the object code - shows a 9938....

 

    I then decided to take a look at the exec source. Uh.. yea... It's much more than a loader for sure.

 

   Looking at all the setup that exec does to create 99/4a environment I can understand why the VDPID code might not work when ran through exec, but @InsaneMultitasker,  let me know if you want to program exec to be VDPID compatible. ?

 

 

 

[+InsaneMultitasker]

28 minutes ago, dhe said:

Looking at all the setup that exec does to create 99/4a environment I can understand why the VDPID code might not work when ran through exec, but @InsaneMultitasker,  let me know if you want to program exec to be VDPID compatible.

Well... that depends on the incompatibility.  I doubt there is anything in Jeff's code that is EXEC-specific.  Did you disable interrupts?  Are the REF'erenced addresses and routines identical (you mention object file+EA, and EXEC is a program that must have been linked with VDP routines)?   I'd post the code in a CPUID/VDPID topic for review if my questions don't elicit a solution.

 

[GDMike]

38 minutes ago, InsaneMultitasker said:

Well... that depends on the incompatibility.  I doubt there is anything in Jeff's code that is EXEC-specific.  Did you disable interrupts?  Are the REF'erenced addresses and routines identical (you mention object file+EA, and EXEC is a program that must have been linked with VDP routines)?   I'd post the code in a CPUID/VDPID topic for review if my questions don't elicit a solution.

 

Chris, Not sure if you need RAM, but there yours if you want, I think it's 24 pieces, since you work on Geneves quite a bit. Otherwise, I'll hang on to them in my stash.

Edited July 20, 2022 by GDMike

[+dhe]

 

>Did you disable interrupts?

Didn't play with interrupts. What ever state is inherited is the state it runs under.

 

>Are the REF'erenced addresses and routines identical?

I didn't ref anything. I used E/A replacement routines that are all inside the code. It's monolithic.

I didn't port this over and run on real hardware, all compiling and testing done under MAME.

[+dhe]

Can someone try this on a real Geneve - loading with exec and then running with gpl + e/a loaded - and let me know the results of each run?

NEOP.tfi

[+9640News]

On my Geneve:

 

CPU:    9995

VDP    V9938

CAL:

 

 

[+dhe]

Was there a difference between running with exec and running with gpl+e/A?

[+dhe]

Even more curious...

 

I ported the code to my geneve:

 

With EXEC:

 

 

 

Then with GPL + E/A

 

[+9640News]

I will doublecheck this evening.  I do not think my eyes played a trick on me on the VDP.

 

I did not run using EXEC.  Missed that request.

 

Beery

[+dhe]

I need to actually look at the vdpid code. Then I think the safest thing to do is make sure I turn off interrupts when those two (CPUID & VDPID) routines are entered.

[+9640News]

With EXEC:

 

CPU:    9995

VDP:   9918A

CAL:

 

With GPL/EA:

 

CPU:    9995

VDP:   V9958

CAL:

 

 

[+InsaneMultitasker]

@dhecan you start a topic (if one doesn't exist already) for the CPUID/VDPID troubleshooting?   If you upload the source there, I'll check EXEC and GPL with respect to the routines you are using. 

 

[+InsaneMultitasker]

Did a bit more work in the SCSI DSR and felt the need to diagram some of the code before I commit to the next low-level, intrusive changes.  For now, I settled on diagrams.net (drawio) for simplicity, file formats, and ability to save work locally or in the cloud.  The screenshot is an excerpt from the SCSI low level routine leading up to calling the read/write IO routines.  (The attached PDF contains the full flow of this section.)  READC and WRITEC are the next on my list, as that is where the caching occurs, and where I think there is a flaw that we haven't knowingly encountered but can occur if two files are written in interleaved fashion.  

 

SCSI-SECTORS Flow Diagram drawio.drawio.pdf

[+InsaneMultitasker]

First attempt at a flow diagram for the OS initialization. Four pages.  Drawio is easy to use but I wish it resized/moved shapes as well as visio.  I'll add the XOP powerups next.

 

 

OS Initialization Draft 20220802.pdf

[+Ksarul]

You could try the Draw app in Libre Office, as that one will at least open Visio documents--but it apparently doesn't have the ability to save in Visio format.

[+InsaneMultitasker]

13 hours ago, Ksarul said:

You could try the Draw app in Libre Office, as that one will at least open Visio documents--but it apparently doesn't have the ability to save in Visio format.

I may give it a try, thanks.  Drawio can export in a variety of formats and yesterday I 'discovered' that Drawio has a desktop version.  So many things to try, so little time...

[+InsaneMultitasker]

MDOS has a feature that allows the user to remap DSK1-to any of the available disk peripheral's devices.  For example, DSK1. could map to a ramdisk, DSK2 to a TIPI DSKx device, DSK3 to the internal ramdisk, and so on. 

 

When the OS starts, DSK1-4 map to the device found at 0x1100 and DSK5-9 are set by the OS to other devices, typically ramdisks.

 

I've been contemplating a minor change whereby REMAPping DSK1-4 from the MDOS command line overrides subsequent automatic detection of these devices during a soft reset (CTRL-ALT-DEL or Ampersand).  A hard reset (OS reload) would still enforce the 0x1100 device detection except when a user modifies the OS via the CYA utility.  DSK5-9 already have this functionality in place.

 

[+InsaneMultitasker]

@9640News - I didn't find a development topic for ABASIC, so I'm posting a note here since Michael asked a question that concerns both ABASIC and the OS. 

 

@mizapf the simple program in the first screenshot displays the extended character set.  In Geneve OS 7.30 (left/first image), some or all of the circled characters are used for the OS version logo.  The second image is the result when I run the same program in MDOS 6.50, with IBMGRF ON.  Are the characters in the second screenshot what you expected to see?

 

  

[+mizapf]

Is IBMGRF equal to codepage 437? As I read in Wikipedia (and remembering the situation back then), Greek beta β and German eszett ß share the same character. This is OK ... at least, we had to live with that in the early days of the PC; so there is nothing to be fixed.

 

 

[+9640News]

The Version LOGO changed characters 228 through 236 from V6.50.

 

We can revert them to their previous definition, but then the LOGO would need to change.

 

Is there a need anywhere for the Greek beta or German eszett or any of the other characters that were modified?

 

 

[+mizapf]

The β / ß is present in both screenshots, as it seems to me (next-to-last row, 6th column). What I meant on my first post is that I did not find it on the keyboard; it would be nice if I could type it.

 

Anyway, this should be kept in the character set, because it does not make sense to offer some of the special characters, but not all.