Page 14 of 15

Re: ADFFS 2.71 beta

Posted: Thu Feb 28, 2019 12:36 am
by Vanfanel
@Jon: I did a long test on Fireball I after updating to latest ADFFS beta, and everything works fine, no more problems! And I was playing for half an hour...

Re: ADFFS 2.71 beta

Posted: Thu Feb 28, 2019 6:33 am
by JonAbbott
Vanfanel wrote: Thu Feb 28, 2019 12:36 am I did a long test on Fireball I after updating to latest ADFFS beta, and everything works fine
Excellent, thanks for confirming.

Re: ADFFS 2.71 beta

Posted: Thu Feb 28, 2019 2:36 pm
by JonAbbott
JonAbbott wrote: Wed Feb 27, 2019 10:46 am Loading StrongEd before ADFFS does appear to cause an issue. It appears to be interfering with the OS_File decryption routine in The Fourth Dimension titles.
2.72j now available, which fixes this issue.

I've rewritten the vector tracking from scratch to track vector claims via OS_Claim / OS_Release. When ADFFS wants to perform IO, it releases the filesystem vectors if they were claimed after ADFFS was loaded, performs its IO and then reinstates after its done.

I tried a few Fourth Dimension titles at random, which all appear to work regardless of the load order of ADFFS. I've also tested a few on RISC OS 3.11 just to be certain.

Re: ADFFS 2.71 beta

Posted: Thu Feb 28, 2019 9:27 pm
by richw
Excellent. I’ve just tried it and confirm that it’s working for me too.

That’s pretty amazing progress, from my initial vague comment to a pretty comprehensive fix in a couple of days!

Re: ADFFS 2.71 beta

Posted: Thu Feb 28, 2019 10:28 pm
by JonAbbott
I'm releasing 2.72 this weekend, after which I'll have very little time to do any further development for probably the rest of the year. Hence the rush to get as many issues resolved as possible.

There's one last issue with 16 colour palettes, which seem to bleed through to 16m colour modes somehow. While testing the Fourth Dimension titles I also noticed some still have flashing colours on the loading screens. I've checked the GraphicsV palette code and can't see any issues, so it may be related to a recent change in RISC OS to resolve flashing colours generating gamma changes.

Re: ADFFS 2.71 beta

Posted: Fri Mar 01, 2019 12:04 am
by Vanfanel
@Jon: So should we update to latest Risc OS ROM now?

Re: ADFFS 2.71 beta

Posted: Fri Mar 01, 2019 8:45 am
by JonAbbott
Vanfanel wrote: Fri Mar 01, 2019 12:04 am should we update to latest Risc OS ROM now?
No need up change RISC OS ROM's, all the RISC OS "features" that have turned up whilst developing 2.71/2.72 have been long running issues and have workarounds implemented, for backward compatibility to RISC OS 3.x

I'm going to roll back the ROM build today, to see if the 16 colour palette issues are new (eg Superpool, Formula Fun). If they're reproducible on older ROM builds, I'll do some last minute debugging to try and track it down and resolve. Apart from that, we're done, so thank you both for all the testing, feedback and the fixes to USBJoystick.

Re: ADFFS 2.71 beta

Posted: Fri Mar 01, 2019 3:11 pm
by JonAbbott
2.72k is available. Whilst looking into the palette issues I spotted a bug in the 26bit Module header check. This build will go to release over the weekend.

I've updated the following packages to resolve minor issues:

F1005601 Break 147 & Superpool (1991) (The Fourth Dimension) (palette issue and missing a directory)
F1031901 Saloon Cars Deluxe (1992) (The Fourth Dimension) (prompted for disc 1)
F1061401 Saloon Cars Deluxe [SA version] (2000) (The Fourth Dimension) (ESC during the intro dropped you to a BASIC commandline)

The palette issue in Superpool was down to using ScreenLoad instead of ADFScreenLoad in the boot script and the palette issue in Formula Fun is due to its reliance on Palette files, which also set the desktop palette.

Re: ADFFS 2.71 beta

Posted: Sat Mar 02, 2019 2:57 am
by Vanfanel
@Jon: Maybe you know it, but Elite hangs the system on quit via CTRL+SHIFT+F12.

Re: ADFFS 2.71 beta

Posted: Sat Mar 02, 2019 9:08 am
by JonAbbott
Vanfanel wrote: Sat Mar 02, 2019 2:57 am Maybe you know it, but Elite hangs the system on quit via CTRL+SHIFT+F12.
I've never tried forcibly terminating Elite as it has is own exit mechanism via F12. Wimp support is a total hack that is known to causes random locks, so I'd expect most if not all Wimp games to lock with CTRL-SHIFT-F12.

The problem with forcibly terminating a Wimp app is the Wimp is in an unknown state, so although the task is terminated when it's next mapped in, the Wimp may lock if it's waiting for a response, or has commands queued for the task. The next time the Wimp tries to poll the app would certainly be a problem, and probably the cause of the lock.

I could probably code something that returns the game to the desktop and then sends a Quit response to its next Wimp_Poll/Wimp_PollIdle, but that's likely to add other issues as the Wimp_Poll would be called with the game now in an unknown state.

I'll add it to the list of things to investigate for the next build.