Page 11 of 15

Re: ADFFS 2.71 beta

Posted: Thu Feb 14, 2019 8:13 pm
by richw
I have tried out a couple of games, and things have been running and exiting quite nicely. So good stuff.

I still have my problem with pointerv and E-Type. I meant to attach this source earlier... It crashes even if there is no code in the pointerv hook!

Re: ADFFS 2.71 beta

Posted: Thu Feb 14, 2019 9:42 pm
by JonAbbott
So close...Talisman quits back to the desktop when it *RMRun's the sound modules from within BASIC. I believe it's occurring because I'm not checking if an Entry point exists for Modules, so it immediately calls OS_Exit - which will return back to the desktop. I'll get that fixed tomorrow.
richw wrote: Thu Feb 14, 2019 8:13 pm I still have my problem with pointerv and E-Type. I meant to attach this source earlier... It crashes even if there is no code in the pointerv hook!
I'll debug it tomorrow and report back.

EDIT: ArcPinball also suffers the RMRun issue

Re: ADFFS 2.71 beta

Posted: Fri Feb 15, 2019 12:24 pm
by JonAbbott
JonAbbott wrote: Thu Feb 14, 2019 9:42 pm So close...Talisman quits back to the desktop when it *RMRun's the sound modules from within BASIC. I believe it's occurring because I'm not checking if an Entry point exists for Modules, so it immediately calls OS_Exit - which will return back to the desktop. I'll get that fixed tomorrow.
...
EDIT: ArcPinball also suffers the RMRun issue
2.72c package now available, which resolved this issue. I've also updated the ArcPinball and Asylum packages, but haven't bumped the version, so remove and reinstall them if you have them installed.
richw wrote: Thu Feb 14, 2019 8:13 pm I still have my problem with pointerv and E-Type. I meant to attach this source earlier... It crashes even if there is no code in the pointerv hook!
God, did I have issues getting it to compile. It seems my DDE is now so out of date, lots of essential things are missing. I had to change various library paths (CLIB:, RISCOSLib:, LocalRes:) and alias out various commands (mkdir, checktoken, Unix.chmod) to get the Ram version built. I couldn't get the debug version to build as it wanted C:Desk.o.Desk_M - which my DDE doesn't contain.

Anyhow...with the Ram build loaded, E-Type was working fine. I tried it with and without a Joystick attached and with *USBJoystick_MouseControl on/off. Incidentally, with my NES controller, the mouse just drifted up/right. That aside, I could control the car in E-Type.

Only advice I can give is to try a clean boot, so nothing else is loaded other than ADFFS and if that doesn't resolve it, try a clean Ram build of USBJoystick, if you've previously been testing the Debug build.

EDIT: I've updated the Break 147 & Superpool package to fix an issue in its !Run, remove and reinstall the package to update.

EDIT2: There's an issue with Burn 'Out which I'm investigating. It's sometimes crashing or hanging when you start a level.

Re: ADFFS 2.71 beta

Posted: Sat Feb 16, 2019 5:08 pm
by JonAbbott
JonAbbott wrote: Fri Feb 15, 2019 12:24 pm There's an issue with Burn 'Out which I'm investigating. It's sometimes crashing or hanging when you start a level.
I've fixed the package and switched it to use the 32bit version of DSymphonyPlayer, remove and reinstall the package if you already have it installed. I've also put the package of the StrongArm version on the release distribution site, as its already freely available from Ian Jeffray's website.

Be aware there's an issue in PackMan which will throw an error when you try to install the game. You can ignore the error and install it anyway.

Re: ADFFS 2.71 beta

Posted: Sat Feb 16, 2019 6:02 pm
by richw
Re. Compiling USBJoystick, sorry, I should have provided some guidance! There was a compiled module in there, and I usually run MkRam to build. My DDE isn't very recent, but you do need OSLib installed. I have never ran MkDebug!

It is odd, because I have little in my boot sequence, and if I rmkill a pointerv-using USBJoystick and load an older one, then it is fine. I feel as if I have narrowed it down to pointerv, but that doesn't really make any sense! I shall try some more methodical testing to see what I can eliminate.

Re: ADFFS 2.71 beta

Posted: Sat Feb 16, 2019 8:49 pm
by JonAbbott
richw wrote: Sat Feb 16, 2019 6:02 pm Re. Compiling USBJoystick, sorry, I should have provided some guidance! There was a compiled module in there, and I usually run MkRam to build. My DDE isn't very recent, but you do need OSLib installed. I have never ran MkDebug!

It is odd, because I have little in my boot sequence, and if I rmkill a pointerv-using USBJoystick and load an older one, then it is fine. I feel as if I have narrowed it down to pointerv, but that doesn't really make any sense! I shall try some more methodical testing to see what I can eliminate.
I've just tried the version in the ZIP and E-Type got stuck loading. It didn't hang, but USB stopped working. So I then replaced the version in ADFFS as I did when I tested previously, and it then worked after a reboot.

If you're reloading USBJoystick after its already loaded, then I think the issue may be that something isn't being released when its RMKilled and it's that that's causing the issues.

Re: ADFFS 2.71 beta

Posted: Sat Feb 16, 2019 9:29 pm
by JonAbbott
2.72d package available. Reboot after downloading as there was a bug in 2.72c which wasn't leaving the Exit/Error handler resident when ADFFS was RMKilled, meaning the machine would likely crash when apps quit/error back to the desktop.

Re: ADFFS 2.71 beta

Posted: Sun Feb 17, 2019 11:32 am
by JonAbbott
I've updated the Command Ship (page zero access fix) and Cycloids (protection failing) packages, please remove and reinstall them to update.

Whilst fixing Cycloids, I took the liberty of removing the "SFX or Music" selection screen and modified it so both SFX and music are active at the same time. It's not perfect as SFX get cut short sometimes, but I think its better than just SFX or music.

There does appear to be an issue with APCS-A based titles (Corruption, The Guild of Thieves, Jinxter), which crash when the Exit handler is called. I suspect it's an issue with CLib 1.01, but haven't investigated yet.

Interesting fact, the packaged games now require 512MB of disc space excluding CD images; with CD images they're just shy of 1.5GB :o. I'm going to have to switch my Pi3 over to SSD and increase the FileCore partition size, as I don't have enough space to store the package sources, the packages and installed versions for testing :roll:.

Re: ADFFS 2.71 beta

Posted: Sun Feb 17, 2019 4:30 pm
by JonAbbott
2.72e package available, which fixes the APCS-A based titles (Corruption, The Guild of Thieves, Jinxter). OS_Control was returning the wrong value in R1 (previous debug buffer pointer)

I've also made some minor changes to the Corruption and Omar Sharif packages, so remove and reinstall them if you have them installed.

Re: ADFFS 2.71 beta

Posted: Sun Feb 17, 2019 11:43 pm
by richw
JonAbbott wrote: Sat Feb 16, 2019 8:49 pm If you're reloading USBJoystick after its already loaded, then I think the issue may be that something isn't being released when its RMKilled and it's that that's causing the issues.
You might be onto something there. I was just trying loading things the other way around, and found the pointerv-less build suddenly made it crash. Maybe that is why... The second-loaded one is wrong? Hmm.

I normally load USBJoystick loads of times, without an issue, so it must be something subtle which only E-Type is showing up. Maybe bytev? I will look.