Home Page Forums MAME MAME Mac weirdness

I’m not overly familiar with the Mac drivers, so I’m not sure if this is me doing something wrong, known limitations, or bugs.

If I try running Mac II booted from a hard disk with two hard disks connected, ADB locks up after the desktop appears. For example if I use this command, the mouse/keyboard won’t respond after the desktop appears:

However, if I boot from a floppy, I can browse both hard disks and copy files between them without ADB locking up. For example, this works:

I occasionally get stuck keys, so I guess this is just ADB HLE weirdness?

Weirder than that though is that System 7.1 doesn’t see some files on the hard disk while System 6 does.

With that previous command (booting from System 6.0.8 on floppy with the System 7.1.1 Pro hard disk attached), I can see the PowerTalk Folder, System 7 Pro Read Me folder, TeachText, etc. as well as any other stuff I add myself.

However, if I boot from System 7 itself, I only see the System Folder in the root folder of the hard disk. For example this only shows the System Folder:

I tried rebuilding the desktop file on the hard disk, but it didn’t improve the situation.

This was all with debug builds, I haven’t tried a release build. Is there something I’m doing wrong?

OK, the missing files in System 7 fixes itself in a release build, so I guess that’s down to a lurking uninitialised variable. ADB locking up with two hard disks when booted from one of them is still present in a release build.

Something is definitely faulty with the ADB HLE, since macpb100 loses the mouse when booting from floppies and loses the keyboard instead when booting from a hard disk.

Osso worked out which uninitialised variables were causing the issues with System 7 filesystem access, so that’s one less thing to worry about.

I’m still getting issues with multiple monitors on emulated Macs. At this point I’m wondering if my PC is haunted. Even if I just use a pair of m2hires cards in a Mac II, it misbehaves. I can move the mouse pointer onto the second monitor or drag a window there, and the OS thinks it’s there, but it isn’t drawn. The second screen just shows the desktop pattern. After dragging it back onto the first screen, (have to do this blind), the contents of the window will be corrupted.

I can confirm a pair of m2hires cards acts weird (the system locks up when I open the Monitor applet, which is definitely bad). A pair of 48gc cards works great though.

ETA: cb264 and 48gc are both "harmless" - I can run a 48gc and 3 cb264s in macii and it works fine.

Just playing around with this on Linux with Mame 0.243.

$ mame macii mac711 -window -ramsize 8M -effect none -video opengl -nomaximize -nounevenstretch -nb9 48gc -nba cb264 -nbb cb264 -nbc cb264 -noautosave

This boots up with 4 monitors like R. Belmont was saying (I assume this is what you meant by "a 48gc and 3 cb264s"). The 48gc card can be put into a color mode with no problem. Mame crashes as soon as I try to put any of the cb264 monitors into a color mode.

Ignoring MAME exception: cb264: unknown video mode 50529027

Fatal error: cb264: unknown video mode 50529027

Didn’t ArBee already fix that in master?

Correct, but they were using 0.243.

Sorry, the Macintosh Display Card is going to have incorrect refresh rates for a while. Calculating the PLL-generated dot clock frequencies is going to take effort.

edit: That was a short while.

video/mac.cpp has some code to force the view depending on the monitor type, which is what causes the weirdness when using a machine with built-in video plus a card. That was considered the best possible solution ~10 years ago to get the right aspect ratio for the full-page display (3:4), can we do something that sucks less now?

I’ve thought about this a bit, and I can’t think of a simple solution yet. Two things that won’t work:

• Using a machine configuration setting for the monitor type, but the time you can read the I/O port to find out what it is, it’s too late to set the screen’s physical aspect ratio. You can’t read I/O ports before the machine starts, but you need to set screens’ physical aspect ratios before layout geometry is calculated.
• If you put the screen in a slot device representing the monitor, its tag changes depending on the monitor type. Slot cards are set up too late to tell device_screen_interface things what the actual screen tags are.

Another annoyance is that there’s no way from within a layout to get a screen’s aspect ratio by tag. That’s still only exposed by index, and screen indices change depend on card setup.

I guess we just leave it for now.

Further complicating the "physical aspect" issue is the existence of rotatable displays with sensors to detect changes during operation between portrait and landscape modes, such as the Facit Twist terminal or (to return to the subject at hand) the Radius Pivot.

But that doesn’t actually change the aspect ratio, just the rotation. The Radius Full Page Pivot still scans along the long axis when it’s vertical. It’s not like the Apple portrait monitors that scan along the short axis.

ArBee: how much explanation for video cards should be included on the wiki page for the 68k Mac drivers? And how much should go in the list, versus having an additional section specifically for video cards?

We're kind of flying blind here in terms of precedent, but I'd at least list what video modes each card supports and any special features (pan/zoom or acceleration).