Compileable version of Alien Breed 3D II - The Killing Grounds

[abu_the_monkey]

It turned out that it was a patch (freewheel) causing the freeze/crash at the credits splash screen on my A4000, however only the AGA screen modes are working for me with the last version posted by paraj.
selecting an RTG screen mode just freezes on the workbench screen, snoopdos gave no insight.
if I have time I will try an earlier version with RTG support on the A4000 cos in my wisdom () I copied over the version I built the other day to test the latest prom paraj and its an arse to get to the cf card in the 'real' machine.

[abu_the_monkey]

I tried the version I built from pippers main repo and again the AGA modes work but the RTG modes crash in workbench before opening a new screen.
this may be due to the state of disrepair my A4000 is in however. It would be nice if someone else with an RTG capable system could test an leave feedback.

[Dunny]

Originally Posted by paraj

Time for some good, old fashioned printf debugging. Please replace tkg_c_dev with the attached (hiresC) executable and try again. It should print some stuff on the screen (waiting a second or so between each update) that can hopefully tell us where it's stuck. It should just show a single line of text while loading.

It's stuck on AB3:SoundFX/Aliens/Meat_Debris. I can upload a video for you if you need it; there's a sequence of files, "in menu" etc before that point but that's where it freezes - if you can call it a freeze, as we already know the flame effect and background scroll continues, I'm assuming, behind the text display.

FWIW, the file is 3,000 bytes dead on, and I've verified that the version of the file on the CF card is the same as the one in the archive I downloaded from Karlos' mod page on github.

[Karlos]

@paraj

Could this be connected to the max sound fx changes?

[paraj]

Originally Posted by abu_the_monkey

I tried the version I built from pippers main repo and again the AGA modes work but the RTG modes crash in workbench before opening a new screen.
this may be due to the state of disrepair my A4000 is in however. It would be nice if someone else with an RTG capable system could test an leave feedback.

It looks like the RTG part is working for Dunny, but unfortunately I can't confirm it myself on real HW (only WinUAE). Do other RTG applications work for you? We could try with a stripped down version or with some debug output.

Originally Posted by Dunny

It's stuck on AB3:SoundFX/Aliens/Meat_Debris. I can upload a video for you if you need it; there's a sequence of files, "in menu" etc before that point but that's where it freezes - if you can call it a freeze, as we already know the flame effect and background scroll continues, I'm assuming, behind the text display.

FWIW, the file is 3,000 bytes dead on, and I've verified that the version of the file on the CF card is the same as the one in the archive I downloaded from Karlos' mod page on github.

That's an unexpected result. No idea what could be causing it. Could you be persuaded to try one more test executable (attached)? It should pinpoint which operation hangs. Just what the last line says is sufficient, but a screenshot would be best.

Originally Posted by Karlos

@paraj
Could this be connected to the max sound fx changes?

Anything is possible, I guess, but I don't see how. What's puzzling is that it's not the first file, it's not compressed, and it's not even the first sound effect!

It works fine for me, even if I run it from a CF card, but then again it's a different setup (KS3.1/AGA only/060).

[tomcat666]

I am running the latest build from the Karlos repository bin dir on pistorm32 and everything is fine. Getting 50 fps on both RTG (c) and AGA (asm) release builds no problem.

[paraj]

Originally Posted by tomcat666

I am running the latest build from the Karlos repository bin dir on pistorm32 and everything is fine. Getting 50 fps on both RTG (c) and AGA (asm) release builds no problem.

That's great, thanks for confirming. My best guess is that it isn't a AB3DII problem, and it's caused by some interaction between CF card driver and PiStorm (tests might show).


That reminds me: @Dunny, could you please also check running from something other than CF card (RAM: would be fine given the amount you have free).

[abu_the_monkey]

It looks like the RTG part is working for Dunny, but unfortunately I can't confirm it myself on real HW (only WinUAE). Do other RTG applications work for you? We could try with a stripped down version or with some debug output.

I am sure it is a memory issue with my voodoo card, changed the workbench screen mode and now I can't get any RTG stuff to work not even workbench.
the card shows up in PCIinfo but any attempt to open a screen and the system either reports that it can't or it dies on its arse.

on the plus side, the AGA version works very well weapons, ammo and health all display correctly

[Dunny]

Originally Posted by paraj

That's an unexpected result. No idea what could be causing it. Could you be persuaded to try one more test executable (attached)? It should pinpoint which operation hangs. Just what the last line says is sufficient, but a screenshot would be best.

Well this one starts up just fine

I'm in and playing. Got to the press enter/* prompt, pressed enter and it's in-game immediately. Sorry but this one may be a bit of a bitch to debug!

[Karlos]

Just posting to say that I'm still reading but completely out of action for a couple of days with other commitments.

This recent issue sounds like a proper Heisenbug.

[paraj]

Originally Posted by Dunny

Well this one starts up just fine

I'm in and playing. Got to the press enter/* prompt, pressed enter and it's in-game immediately. Sorry but this one may be a bit of a bitch to debug!

Problem solved. We'll just ship that version then


One more. This one should get stuck, but perhaps you can make it progress by holding down the left mouse button. If that's the case does releasing the LMB make it get stuck again? Again the interesting thing is what operation was last reported in case it got stuck (expecting something like e.g. "T AB3:...." or "ExamineFH")

[Karlos]

"it works on my machine" is how docker got started...

[saimon69]

Originally Posted by Karlos

"it works on my machine" is how docker got started...

Linux users know it as WorksForMe[TM]

[Dunny]

Originally Posted by saimon69

Linux users know it as WorksForMe[TM]

We use LLMs extensively at work now. While testing, a colleague asked for some python code, which was obviously wrong. When asked why the LLM thought it might be wrong, it replied with "works here".

[Dunny]

Originally Posted by paraj

Problem solved. We'll just ship that version then


One more. This one should get stuck, but perhaps you can make it progress by holding down the left mouse button. If that's the case does releasing the LMB make it get stuck again? Again the interesting thing is what operation was last reported in case it got stuck (expecting something like e.g. "T AB3:...." or "ExamineFH")

This one also runs fine. Interestingly it can get stuck at random points when running from CF Card, so I guess it's likely that which is causing the issues - I suspect a PiStorm issue as they were only recently fixed.

Apologies for the run-around guys.

[Karlos]

@paraj

PR looks good. The only suggestion I have - and it's not a blocker - is to consider using the dos.library functions for console output. If it has any meaningful impact on executable size, anyway.

[Karlos]

@paraj

Merged! The C stdio v dos.library can be a future consideration.

[Karlos]

Originally Posted by dunny

we use llms extensively at work now. While testing, a colleague asked for some python code, which was obviously wrong. When asked why the llm thought it might be wrong, it replied with "works here".

Quote Of The Day

[paraj]

Originally Posted by Dunny

This one also runs fine. Interestingly it can get stuck at random points when running from CF Card, so I guess it's likely that which is causing the issues - I suspect a PiStorm issue as they were only recently fixed.

Apologies for the run-around guys.

Where did it get stuck, and did LMB help? Even if it only shows on your pistorm from CF card for now, it could still help debugging other issues later (on pistorm/in AB3DII or other projects). LMB disables the blitter effect (fire) temporarily, and my suspicious is that having that effect active could perhaps cause the device/filesystem process to have issues (be starved/miss deadline/etc).



The code should be fine, but maybe it could be made more friendly by e.g. lowering priority of the blitter task, or indicate that we should work on converting that effect to use the CPU.

Originally Posted by Karlos

@paraj

PR looks good. The only suggestion I have - and it's not a blocker - is to consider using the dos.library functions for console output. If it has any meaningful impact on executable size, anyway.

The stdio usage (printf) was already there, I just converted it to the more lightweight puts that also adds a newline. Checking the code I can see that it's the only console output in the whole executable.


Really that message should be ditched all together. It'll only show up if the player pressed "cancel" in the screen mode selection dialog in which case we should just exit.


Also reviewing my own changes, I think the use of an Alert should be changed to just a normal requester with an "exit" button. Normal people will be starting the game from the WB, and shouldn't be shown a console window like they're some kind of dirty developer.


I also still think trying to automatically open assets from the current directory and/or PROGDIR: if Open() fails is a good idea (so we can do away with the need for the startup script for plain old TKG).


I'll look at both of those things, unless y'all disagree.

Originally Posted by Karlos

Quote Of The Day

"ChatGPT are you lying to me" "No...."

[ZyBeR]

Originally Posted by paraj

LMB disables the blitter effect (fire) temporarily, and my suspicious is that having that effect active could perhaps cause the device/filesystem process to have issues (be starved/miss deadline/etc).

The code should be fine, but maybe it could be made more friendly by e.g. lowering priority of the blitter task, or indicate that we should work on converting that effect to use the CPU.

Maybe it's just me, but does that effect really add anything to the game?
I would argue it better to remove it and simplify the whole menu implementation.