ADFFS 2.63 beta

Discuss ADFFS development versions and upcoming additions
Vanfanel
Posts: 373
Joined: Mon Sep 16, 2013 12:01 am

Re: ADFFS 2.63 beta

Post by Vanfanel » Mon Oct 16, 2017 11:03 am

@Jon: How can I find out the correct partition number? Is there a RISC OS console command to list partition numbers?

JonAbbott
Posts: 1736
Joined: Thu Apr 11, 2013 12:13 pm
Location: Essex

Re: ADFFS 2.63 beta

Post by JonAbbott » Mon Oct 16, 2017 11:57 am

Vanfanel wrote:@Jon: How can I find out the correct partition number? Is there a RISC OS console command to list partition numbers?
Plug it into a PC and look in Windows Disk Manager or use a Linux equivalent, RISC OS doesn't understand partitions.
JonAbbott wrote:There is an issue affecting Zalaga, but I believe that's specific to Zalaga and does not affect any other games.
I've tracked this problem down, !65Host is attempting to directly write to the OS_Byte variable table. I'll probably botch it for the time being as its not going to be straight forward implementing a Page Zero read/write translation to the equivalent OS_Byte call.

Vanfanel
Posts: 373
Joined: Mon Sep 16, 2013 12:01 am

Re: ADFFS 2.63 beta

Post by Vanfanel » Mon Oct 16, 2017 3:34 pm

@Jon: the reported regressions are fixed now (Stunt Racer 2000 showing an error on exit is still there but I think it was there before).
About HMM2, I adjusted the MountDosPartition script, since FAT32 partition number is 2 according to GNU/Linux fdisk, and now the partition is mounted well.
However, after it has been mounted, I double click on CDFaker, then on !Heroes2, and I get "Could not open filesystem" error. I am almost there I guess, but there's something I am still doing wrong...

I also tried cd-ing into the fat32fs, and then manually:
cdfaker_mount 0 hmm2/cue

But I get "Not an ISOimage or CUE file"

JonAbbott
Posts: 1736
Joined: Thu Apr 11, 2013 12:13 pm
Location: Essex

Re: ADFFS 2.63 beta

Post by JonAbbott » Mon Oct 16, 2017 5:07 pm

JonAbbott wrote:I've tracked this problem down, !65Host is attempting to directly write to the OS_Byte variable table. I'll probably botch it for the time being as its not going to be straight forward implementing a Page Zero read/write translation to the equivalent OS_Byte call.
Updated Modules on the dev site which hopefully get Zalaga working. I've botched the OS_Byte variable table issue (in that writes to it are ignored) and have modified the shutdown process so the BBC Emulator is killed.
Vanfanel wrote:after it has been mounted, I double click on CDFaker, then on !Heroes2, and I get "Could not open filesystem" error
You forgot to update the path on the CDFaker_Mount line, it needs to be the full path including the FAT32 partition name.
Vanfanel wrote:I get "Not an ISOimage or CUE file"
You need to set the filetype of the /CUE file to "CUE" and the /BIN file should be "Data"

Vanfanel
Posts: 373
Joined: Mon Sep 16, 2013 12:01 am

Re: ADFFS 2.63 beta

Post by Vanfanel » Mon Oct 16, 2017 11:48 pm

@Jon: Got it to work finally. Yeah. it was the partition name in !Heroes2 run file, and the filetypes not set :)
What a nice game HMM2 was, it's an awesome adition to the already incredible Risc OS Pi library ADFFS is bringing back to life.
However, it seems to have problems reading music: the cursor becomes very "sticky" when the SD card access light comes up. No way around it? (Keeping the great audioCD OST) We have SO much RAM on the Pi2/3 that almost all the CD could be in cache :lol:

JonAbbott
Posts: 1736
Joined: Thu Apr 11, 2013 12:13 pm
Location: Essex

Re: ADFFS 2.63 beta

Post by JonAbbott » Tue Oct 17, 2017 12:39 pm

Vanfanel wrote:However, it seems to have problems reading music: the cursor becomes very "sticky" when the SD card access light comes up. No way around it? (Keeping the great audioCD OST) We have SO much RAM on the Pi2/3 that almost all the CD could be in cache :lol:
As I warned you, I've yet to optimize the audio buffering in CDFaker. I wouldn't expect an SD to cause buffering issues though, what speed is your SD?

Vanfanel
Posts: 373
Joined: Mon Sep 16, 2013 12:01 am

Re: ADFFS 2.63 beta

Post by Vanfanel » Tue Oct 17, 2017 2:24 pm

@jon: its a class 4 trascend micro SD on a Pi2.
I tried HMM2 on a Pi1 and mouse cursor is not SO sticky when reading music from SD as it is on Pi2/3!
SD access on Risc OS seems to be slower on Pi2/3 than on Pi1. I had already seen that on Doom.

JonAbbott
Posts: 1736
Joined: Thu Apr 11, 2013 12:13 pm
Location: Essex

Re: ADFFS 2.63 beta

Post by JonAbbott » Tue Oct 17, 2017 9:42 pm

I've play tested it for hours on a Pi3 with no noticeable mouse problems, I'll try Pi1 and Pi2 when I get around to building them for the London Show

Vanfanel
Posts: 373
Joined: Mon Sep 16, 2013 12:01 am

Re: ADFFS 2.63 beta

Post by Vanfanel » Wed Oct 18, 2017 1:54 am

@jon: try to move the mouse in circles, doing it on the main menu will do.
What's the SD card you are using to store the CUE/BIN? Brand, class...

JonAbbott
Posts: 1736
Joined: Thu Apr 11, 2013 12:13 pm
Location: Essex

Re: ADFFS 2.63 beta

Post by JonAbbott » Wed Oct 18, 2017 10:00 am

New build up on the dev site, I found some fairly fatal bugs whilst testing on the Iyonix. There's still some lurking, as some games are still failing.
Vanfanel wrote:@jon: try to move the mouse in circles, doing it on the main menu will do.
What's the SD card you are using to store the CUE/BIN? Brand, class...
I have a class U1 which is 10MB/s. Its the speed for small random access reads that's going to impact playing audio on an SD.

I've just retested doing circles on the main menu, I do see a slight pause when it buffers more audio but its probably not something I can resolve. I could try reading smaller chunks more often, but that will probably make it worse as the pauses will just happen more often. The root of the issue is the fact you can't do background transfers from disc under RISC OS.

I do plan to add parameters to configure the buffer size, initial fill size and read-ahead size which will resolve the problem if the buffer is big enough to read the whole track. It would still suffer pauses whilst it does the read-ahead though, or a massive pause when a track starts, if the read-ahead is set large enough to pre-cache the whole track.

Post Reply