TIPI Usage and Support

[+jedimatt42]

It was true a month ago, and it is true today, that there is no intention to update the ROM in the future.

[+arcadeshopper]

4 hours ago, mizapf said:

If you want to use the new DSR in the MAME emulation, replace the file ti99_tipi.zip in your rompath with the attached file.

 

However ... due to the nature of MAME, we will get checksum errors each time the ROMs are changed, unless I also change the checksums inside MAME for the TIPI ROM. Yes, this is intended and a Good Thing^TM.

 

These error messages can be ignored, but I think we cannot turn them off completely (this is, IIRC, also on purpose from the core devs).

 

Does it make sense to update the checksums now, or do we expect more frequent changes in the DSR in the future? If so, I'd update the checksums in longer terms (every 3rd time or so).

 

When I update them, users of the old DSR will get checksum errors, of course.

ti99_tipi.zip 1.91 kB · 1 download

@jedimatt42 can answer better but i don't think there will be another rom update for a good while, the last one was over a year ago.

 

 

[+Atari2600PAL]

When booting p-code card, if the TIPI-PEB is at CRU 0x1000 (FDC at 0x1100) the p-code card doesn't attempt any access of the physical drives even when there are no TIPI drive mappings

 

If TIPI CRU is moved to 0x1800 then p-code card works as expected

 

The TIPI log contains only the following when set at 0x1000 regardless of any actions attempted on the p-code card:

 

2023-08-14 19:43:12,471 TipiService : INFO     physical mode enabled

2023-08-14 19:43:12,480 TipiService : INFO     TIPI Ready

2023-08-14 19:45:11,911 LevelTwo    : INFO     sector request

2023-08-14 19:45:11,913 LevelTwo    : INFO     unit: 1, sector: 4, read: 1

2023-08-14 19:45:11,914 LevelTwo    : INFO     no drive mapped for unit 1

2023-08-14 19:45:24,218 LevelTwo    : INFO     sector request

2023-08-14 19:45:24,220 LevelTwo    : INFO     unit: 1, sector: 4, read: 1

2023-08-14 19:45:24,220 LevelTwo    : INFO     no drive mapped for unit 1

2023-08-14 19:45:24,226 LevelTwo    : INFO     sector request

2023-08-14 19:45:24,228 LevelTwo    : INFO     unit: 2, sector: 4, read: 1

2023-08-14 19:45:24,228 LevelTwo    : INFO     no drive mapped for unit 2

2023-08-14 19:45:24,242 LevelTwo    : INFO     sector request

2023-08-14 19:45:24,244 LevelTwo    : INFO     unit: 3, sector: 4, read: 1

2023-08-14 19:45:24,244 LevelTwo    : INFO     no drive mapped for unit 3

 

[+jedimatt42]

Yes, p-code with TIPI only really works if there is no other disk controller and TIPI is at CRU base  1100.

 

There is nothing we can do to fix it short of fixing the p-code system.

[+arcadeshopper]

2 hours ago, jedimatt42 said:

Yes, p-code with TIPI only really works if there is no other disk controller and TIPI is at CRU base  1100.

 

There is nothing we can do to fix it short of fixing the p-code system.

shouldn't it ignore it if tipi is at 1000 and there's no disks mapped?

 

no because the psystem grabs the first controller it sees.. so tipi needs to be > 1100 

 

[+Atari2600PAL]

3 hours ago, jedimatt42 said:

Yes, p-code with TIPI only really works if there is no other disk controller and TIPI is at CRU base  1100.

 

There is nothing we can do to fix it short of fixing the p-code system.

Thanks. With the extra disk mappings you've added I could just keep it at 1800 permanently, but will fit a switch as there are times when I'll want to use TIPI for dsk1

 

1 hour ago, arcadeshopper said:

shouldn't it ignore it if tipi is at 1000 and there's no disks mapped?

 

no because the psystem grabs the first controller it sees.. so tipi needs to be > 1100 

 

Thanks, definitely need to fit a switch then

 

Makes me wonder if there's a way to wire the TIPI CRU jumper to the p-code card on/off switch, but that's beyond me

[+jedimatt42]

8 hours ago, arcadeshopper said:

shouldn't it ignore it if tipi is at 1000 and there's no disks mapped?

 

no because the psystem grabs the first controller it sees.. so tipi needs to be > 1100 

 

More precisely, p-code is a different operating system. The feature that allows TIPI to skip requests for an unmapped drive is a part of the third, high level, file access APIs of the console DSRLNK, and most clones. 

 

P-code uses the lowest level API, sector read/write. And when it was created, nothing had DSRs that 'skipped' a request and so the implementation of p-code didn't consistently allow for the feature. Worse yet, some pieces of p-code assume the floppy controller is at CRU base 1100 and optimize out the legacy DSRLNK altogether.

 

Supporting p-code stopped being my intention. Even under 1st party control, TI didn't protect the hardware abstraction layer correctly.

 

The sector access support I implemented is good enough for Plato, and a few poorly written 4A utilities. But even for those, generally only if TIPI is the sole disk provider.

[+mizapf]

1 hour ago, jedimatt42 said:

P-code uses the lowest level API, sector read/write. And when it was created, nothing had DSRs that 'skipped' a request and so the implementation of p-code didn't consistently allow for the feature. Worse yet, some pieces of p-code assume the floppy controller is at CRU base 1100 and optimize out the legacy DSRLNK altogether.

Why does this always sound like different working groups in the same company that don't really talk with each other or even understand what the others are doing ...

[+arcadeshopper]

1 hour ago, mizapf said:

Why does this always sound like different working groups in the same company that don't really talk with each other or even understand what the others are doing ...

because that's TI in a nutshell

[+dhe]

 >Why does this always sound like different working groups in the same company that don't really talk with each other or even understand what the others are doing ...

 

 I'm not sure if true with the pSystem, but TI farmed out a lot of development - especially to SMU - to Universities.

 

 

[+mizapf]

I just got my second Raspberry Pi 4 ... finally. It's the 8 GiB version which I'll use in particular to build MAME on (the 4 GiB runs out of memory and gets into heavy thrashing).

 

So I can do some reshuffle at home; the previous Raspi 4 will replace the current Raspi 3, and that one could take the role of the TIPI Raspi. I suppose a Raspi 3 is enough for TIPI, right?

[+arcadeshopper]

1 hour ago, mizapf said:

I just got my second Raspberry Pi 4 ... finally. It's the 8 GiB version which I'll use in particular to build MAME on (the 4 GiB runs out of memory and gets into heavy thrashing).

 

So I can do some reshuffle at home; the previous Raspi 4 will replace the current Raspi 3, and that one could take the role of the TIPI Raspi. I suppose a Raspi 3 is enough for TIPI, right?

right 

[+Atari2600PAL]

On 8/15/2023 at 7:20 AM, Atari2600PAL said:

Thanks, definitely need to fit a switch then

Just fitted a switch and it works a treat

 

Is it ok to flip the switch while PI is running, but PEB is powered off please?

 

Many thanks 

[+arcadeshopper]

1 hour ago, Atari2600PAL said:

Just fitted a switch and it works a treat

 

Is it ok to flip the switch while PI is running, but PEB is powered off please?

 

Many thanks 

Yes, changing cru has nothing to do with the PI..  make sure you switch with the PEB off though. 

[+Atari2600PAL]

1 minute ago, arcadeshopper said:

Yes, changing cru has nothing to do with the PI..  make sure you switch with the PEB off though. 

Many thanks @arcadeshopper

[+jedimatt42]

I've dropped an update to ForceCommand that supports the CS1 and DSK5-9 mapping with the builtin TIPIMAP command. 

[+jedimatt42]

Update 3.21 - 2023-08-20

 

- Fix TipiWatchdog.service - restore functionality that tipi.service restarts when PI detects reset signal from TIPI hardware. I think this has been broken since the Le Potato alternative PI function went in. This is also a good example of why whitespace should not be a substitute for { } in a programming language. As you can see here, the change is just deleting 6 space characters: https://github.com/jedimatt42/tipi/commit/6d7de225b1c67f252aaa1e02d1608838295ed9c3

[+OLD CS1]

43 minutes ago, jedimatt42 said:

This is also a good example of why whitespace should not be a substitute for { } in a programming language.

So, Python uses indentation as a grouping delimiter?  What is this, COBOL??

[+Vorticon]

Happy to report that the CS1 functionality seems to work. I tested it using Video Chess to save a game and I am able to save and load from a folder without issues. This is actually great because there are times when I have to interrupt an ongoing game and previously I would just let it go rather than pull out the old cassette recorder and go through the motions. Now it's painless and takes literally a second for the save/load function to complete.

Nice work!

 

[+mizapf]

19 hours ago, OLD CS1 said:

So, Python uses indentation as a grouping delimiter?  What is this, COBOL??

Yes, the interpreter complains about an invalid indentation if you happen to hit the space bar once more.

[Geoff Oltmans]

On 8/20/2023 at 12:05 PM, jedimatt42 said:

Update 3.21 - 2023-08-20

 

- Fix TipiWatchdog.service - restore functionality that tipi.service restarts when PI detects reset signal from TIPI hardware. I think this has been broken since the Le Potato alternative PI function went in. This is also a good example of why whitespace should not be a substitute for { } in a programming language. As you can see here, the change is just deleting 6 space characters: https://github.com/jedimatt42/tipi/commit/6d7de225b1c67f252aaa1e02d1608838295ed9c3

Ugh, that's been my life lately, don't get me started. My favorite is when pro-whitespace people tell you to just "turn on visible whitespace in your editor." Uh, how is that any better than already visible scope markers like {}???

[+Atari2600PAL]

CS1 Support is working fine for me (thanks for the updated DSR @arcadeshopper)

 

I did mess up initially and started my path with TIPI. (TIPI.DISKS.TESTCS1 for example) and got I/o error 66, but once I changed that to just DISKS.TESTCS1 it worked great

 

Many thanks

[Lshumphries]

I recently purchased a TIPI/32k that fits in the speech synthesizer box.  It came with an as card with tipi sdimage version 2.5.  Initially I bridged power from the TIPI board to the Pi board.  All seemed well, as I was able to “Call tipi” and configure WiFi.  When I tried to “call tipi(“tipi.net.telnet”), however, it seems that the entire computer reset.  I thought that maybe too much power was being pulled from the console.  To begin ruling things out, I powered the PI zero separately.  Unfortunate that did not fix the issue. I was able to confirm that I could ssh with PuTTY and call the commands to expand storage on the so card.  I am not sure what is going on and was hoping for some advice.  Has anyone seen this type of issue before? Before upgrading to the new release(I believe 3.17), I wanted to see if anyone else may have seen this and could offer some assistance.  Any thoughts on what may be happening.

[+Atari2600PAL]

Have you tried with alpha lock on?

 

That works for me, but in lowercase system resets

[Lshumphries]

54 minutes ago, Atari2600PAL said:

Have you tried with alpha lock on?

 

That works for me, but in lowercase system resets

Alpha lock for the “TIPI.NET.TELNET”?  I will try that.

 

That was it!

Edited August 26, 2023 by Lshumphries