NEW EXPERIEMNT
JDW, I went back and read your testing methods when I first proposed the "magic" system. (I love that name BTW).
After reading your latest with the HD20, I started thinking about why it was necessary to install an AT printer driver from system 3.2.
Here's what I came up with, the one thing we overlooked.
I took a peek inside the HD20 INIT with Resedit and discovered that it was basically a resource called PTCH and contained 3 files. Basically the ROM "patch". Of the three files the largest was "TFS" which I took to be "everything a 64K ROM needed to know to read and write HFS". Then significantly smaller was ".Sony" which is the disk driver already in every ROM, presumably updated to include 800K drives and the HD20 hard disk. Finally at a mere 290 bytes, something called a "Dispatch Kernel" about which I could not speculate. So I deleted "TFS" and the "Dispatch Kernel". Call me impulsive, but ...
Next I replaced this modified HD20 INIT with the clean HD20 INIT in our "magic system" 2.0/4.1 and guess what, it booted up (with no Welcome screen message at all)! But it would not recognize HFS disks, only the 800K drives. That really got me thinking about version numbers. Since the Finder went up to 5.0 in the original HFS HD20 INIT system and we think 4.1 Finder is the reason the 800K disks are formatted MFS instead of HFS in the magic system, I noticed that the first official HFS system was 3.0 when the Mac Plus was released. The HD20 was released with system 2.1. So I put a clean System 2.1 and Finder 4.1 on a disk with the modified HD20 INIT and it worked fine, just like the "magic system".
http://FYI, In order to format 800K MFS disks, you must used the MODIFIED HD20 INIT, otherwise it acts just as the regular HFS System. However, even though it reads HFS this way, it does not seem to write it. Folders written to a known HFS disk under this system do not show up correctly under an HFS system.
Now I need others to confirm this before I get too excited.
Here's what I think though. I think the reason JDW's HD20 doesn't start up with the magic disk is because system 2.0 as he says doesn't know what to do with it. I'd like to see if 2.1 understands it. In your case, you should take the original HD20 startup disk with System 2.1/Finder 5.0 and replace the 5.0 finder with clean 4.1 Finder and see if your HD20 will boot from that. If not, it's a fair bet what the HD20 needs is in the 5.0+ Finder. At least that would solve that question. Now, I agree with you that the system folder on your HD20 does not directly cause the crash you experience ... but indirectly, it just being there might. Keep in mind the HD20 INIT is designed to not only make 64K ROMs HFS, 800K & HD20 aware, but also to look for a system on the HD20 drive and hand off to it and eject the HD20 Startup disk. Most likely some of that is handled by the System itself. If your HD20 did not have a system folder, I wonder if the magic system would have mounted it normally. Alternately, since 2.0 is an MFS system, I wonder if it could actually even see your system inside its System Folder. You might try using magic 2.0/4.1 with the System and Finder loose in drive and see if the crash is the same.
If 2.1/4.1 works then try the bare HD20 and see what happens. Maybe even a modified bare HD20 with the .Sony driver and the "Dispatch Kernel", which may be the bit that hands off startup to the hard drive.
Unless I find a problem using the modified HD20 INIT, it makes the most sense to use, as it will save you about 24K over the current HD20 INIT. Also, it prevents the 64K ROM from reading HFS disks, possibly causing data corruption.
What it does tell me though, is something was added to System 2.1 that tells it to look for the HD20 INIT, that's why it won't work with a clean System 2.0, the System has to know to look for a patch as well as which patch. In this case it must be told to look for "Hard Disk 20", which by the way cannot be re-named like the System & Finder. I also think whatever was added to 2.1 is not the same as what I suspect is the complete HFS overhaul represented by System 3.0. in other words, possibly a quick fix to get the HD20 out the door. (BTW, until I read some marketing info, I did not realize the HD20 had been announced in April '85 but did not ship until September! If what I suspect is true, no wonder they slapped a patch on it and as I've read and did not make much fanfare about HFS when they released it – 'cause it wasn't ready for general release!) The only thing I can think of for why the System 3.2 AT ImageWriter install trick works is that it needs certain resources it also shares with the HD20.
What I did not mention here is that the .Sony driver ONLY modified version of the HD20 still will not load on the 128K, which puzzles me since the driver info alone should not add significantly to the System heap enough to keep it from loading into available RAM. That tells me that whatever was added to System 2.1 also specifically instructed the HD20 NOT to load when only 128K RAM was present. This gives me hope for finding and eliminating that code to at least Make the 128K compatible with 800K drives.
To summarize: Possible new way to Format 800K MFS disks with 64K ROMS/
Clean System 2.1
Clean Finder 4.1
MODIFIED HD20 INIT ("TFS" resource removed with ResEdit)
No magic.
JDW, I went back and read your testing methods when I first proposed the "magic" system. (I love that name BTW).
After reading your latest with the HD20, I started thinking about why it was necessary to install an AT printer driver from system 3.2.
Here's what I came up with, the one thing we overlooked.
I took a peek inside the HD20 INIT with Resedit and discovered that it was basically a resource called PTCH and contained 3 files. Basically the ROM "patch". Of the three files the largest was "TFS" which I took to be "everything a 64K ROM needed to know to read and write HFS". Then significantly smaller was ".Sony" which is the disk driver already in every ROM, presumably updated to include 800K drives and the HD20 hard disk. Finally at a mere 290 bytes, something called a "Dispatch Kernel" about which I could not speculate. So I deleted "TFS" and the "Dispatch Kernel". Call me impulsive, but ...
Next I replaced this modified HD20 INIT with the clean HD20 INIT in our "magic system" 2.0/4.1 and guess what, it booted up (with no Welcome screen message at all)! But it would not recognize HFS disks, only the 800K drives. That really got me thinking about version numbers. Since the Finder went up to 5.0 in the original HFS HD20 INIT system and we think 4.1 Finder is the reason the 800K disks are formatted MFS instead of HFS in the magic system, I noticed that the first official HFS system was 3.0 when the Mac Plus was released. The HD20 was released with system 2.1. So I put a clean System 2.1 and Finder 4.1 on a disk with the modified HD20 INIT and it worked fine, just like the "magic system".
http://FYI, In order to format 800K MFS disks, you must used the MODIFIED HD20 INIT, otherwise it acts just as the regular HFS System. However, even though it reads HFS this way, it does not seem to write it. Folders written to a known HFS disk under this system do not show up correctly under an HFS system.
Now I need others to confirm this before I get too excited.
Here's what I think though. I think the reason JDW's HD20 doesn't start up with the magic disk is because system 2.0 as he says doesn't know what to do with it. I'd like to see if 2.1 understands it. In your case, you should take the original HD20 startup disk with System 2.1/Finder 5.0 and replace the 5.0 finder with clean 4.1 Finder and see if your HD20 will boot from that. If not, it's a fair bet what the HD20 needs is in the 5.0+ Finder. At least that would solve that question. Now, I agree with you that the system folder on your HD20 does not directly cause the crash you experience ... but indirectly, it just being there might. Keep in mind the HD20 INIT is designed to not only make 64K ROMs HFS, 800K & HD20 aware, but also to look for a system on the HD20 drive and hand off to it and eject the HD20 Startup disk. Most likely some of that is handled by the System itself. If your HD20 did not have a system folder, I wonder if the magic system would have mounted it normally. Alternately, since 2.0 is an MFS system, I wonder if it could actually even see your system inside its System Folder. You might try using magic 2.0/4.1 with the System and Finder loose in drive and see if the crash is the same.
If 2.1/4.1 works then try the bare HD20 and see what happens. Maybe even a modified bare HD20 with the .Sony driver and the "Dispatch Kernel", which may be the bit that hands off startup to the hard drive.
Unless I find a problem using the modified HD20 INIT, it makes the most sense to use, as it will save you about 24K over the current HD20 INIT. Also, it prevents the 64K ROM from reading HFS disks, possibly causing data corruption.
What it does tell me though, is something was added to System 2.1 that tells it to look for the HD20 INIT, that's why it won't work with a clean System 2.0, the System has to know to look for a patch as well as which patch. In this case it must be told to look for "Hard Disk 20", which by the way cannot be re-named like the System & Finder. I also think whatever was added to 2.1 is not the same as what I suspect is the complete HFS overhaul represented by System 3.0. in other words, possibly a quick fix to get the HD20 out the door. (BTW, until I read some marketing info, I did not realize the HD20 had been announced in April '85 but did not ship until September! If what I suspect is true, no wonder they slapped a patch on it and as I've read and did not make much fanfare about HFS when they released it – 'cause it wasn't ready for general release!) The only thing I can think of for why the System 3.2 AT ImageWriter install trick works is that it needs certain resources it also shares with the HD20.
What I did not mention here is that the .Sony driver ONLY modified version of the HD20 still will not load on the 128K, which puzzles me since the driver info alone should not add significantly to the System heap enough to keep it from loading into available RAM. That tells me that whatever was added to System 2.1 also specifically instructed the HD20 NOT to load when only 128K RAM was present. This gives me hope for finding and eliminating that code to at least Make the 128K compatible with 800K drives.
To summarize: Possible new way to Format 800K MFS disks with 64K ROMS/
Clean System 2.1
Clean Finder 4.1
MODIFIED HD20 INIT ("TFS" resource removed with ResEdit)
No magic.