CompuServe Thread

Forum unknown · Help/Announcements

#uploading

17 messages in this thread
#34355From: Don Curtis/SYSOPSep 25, 1986 10:08 PM
Bela, I'm guessing that OnLine! is able to recover from an error during the transfer while Ahost can't so that explains the no-crash vs crash. As far as the explicit call to CIS — back a few years ago when B Protocol became available for the Atari 800 series, they had the same problem…so I do suspect CIS may have some fault in the bad uploads. Don
#34521From: Bela LubkinSep 27, 1986 7:26 PM
If they put it in ROM, they will surely allow you to boot KickStart off a floppy as well. That's just common sense. However, the idea of having a ROM as well as the KS RAM is new to me. 256K of RAM is not a big expense, but I'm not sure about the circuitry that allows it to be write-protected. (Seems pretty simple, but what do I know). So maybe they've gotten rid of the WCS daughterboard, and just added another 256K to the standard memory, with the write-protect hardware, and the ROM. Yes, then you could boot ROM to WCS RAM, and then as the disk-resident part loaded, it could patch the WCS copy of the ROM as necessary. A neat way to have patchable ROM. Or how about this: 256K of battery-backed RAM. No ROM. You have to use the KickStart disk the first time you boot the machine (unless it's done at the factory), but then it isn't needed again until you get a new version of KS. Now that would be >nice<… – Bela
#34536From: Don Curtis/SYSOPSep 27, 1986 9:32 PM
Bela, Aren't guessing games fun! <grin> I was speculating, and I do see one obvious problem….the daughter board takes up one of the ROM sockets (which socket I presume would be needed for the OS ROMs) so we are really left with the ST style OS, you can have it in RAM, but lose 256K or you can have it in ROM and have full memory. Even with the ST's OS (TOS) in ROM, you can boot up the TOS disk and load the OS from disk…but do lose the RAM. One other somewhat obvious solution is to not offer the ROM upgrade to A1000 owners, but keep that OS on disk and put the ROM's only in upgrade machines. I mean Hazy did say "Not what everyone thinks……….. Don
#34818From: Dave Haynie/C128 LandSep 30, 1986 11:11 PM
There's no simple way to have both ROM and WCS RAM in an A1000 board. And they've done a revised A1000 board that has WCS down on the main board, and no ROM slots. I don't know if that's what's actually shipping, but if it is that would really close the issue for a widespread modification of A1000 machines to use ROM instead of RAM. The ROM is likely to appear in a future machine; possibly all future machines. And there's been a proposal to also sell a WCS card as an add on for those with a new machine that want WCS instead of ROM. Though I'm sure any new Amiga system will have more than the 256K base of the A1000, especially since we've pretty much established 512K as a minimal setup for any useful multitasking. But unlike our friends over in ST land, its a small matter to replace alot of the ROM Kernal routines with disk loaded routines. I know that this can be done on a library-by-library basis, so any interm upgrades between major ROM releases could be via a few libraries, not the whole 256K of ROM Kernal. Any major revision COULD be upgraded with a new set of ROMs. The thing I don't like about a ROM locked ROM Kernal, however, is that it makes a software upgrade more difficult to achieve. If we make the 1.2 workbench and kickstart disks free upgrades, or even charge for cost of materials, most everyone will go out, pop in the new Kickstart, and be on their way. And with the overwhelming prevalence of 1.2 over 1.1 (hasn't happened yet, waiting on that FINAL release…), 1.1 becomes a thing of the past. If everyone had to open their machines and swap ROMs, it would take much longer for updates to be globally in place, and the percentage of holdouts could be large enough to form a critical mass. At this point, developers would write for both systems, avoiding any enhancements in the new system. -Hazy
#34818From: Dave Haynie/C128 LandSep 30, 1986 11:11 PM
There's no simple way to have both ROM and WCS RAM in an A1000 board. And they've done a revised A1000 board that has WCS down on the main board, and no ROM slots. I don't know if that's what's actually shipping, but if it is that would really close the issue for a widespread modification of A1000 machines to use ROM instead of RAM. The ROM is likely to appear in a future machine; possibly all future machines. And there's been a proposal to also sell a WCS card as an add on for those with a new machine that want WCS instead of ROM. Though I'm sure any new Amiga system will have more than the 256K base of the A1000, especially since we've pretty much established 512K as a minimal setup for any useful multitasking. But unlike our friends over in ST land, its a small matter to replace alot of the ROM Kernal routines with disk loaded routines. I know that this can be done on a library-by-library basis, so any interm upgrades between major ROM releases could be via a few libraries, not the whole 256K of ROM Kernal. Any major revision COULD be upgraded with a new set of ROMs. The thing I don't like about a ROM locked ROM Kernal, however, is that it makes a software upgrade more difficult to achieve. If we make the 1.2 workbench and kickstart disks free upgrades, or even charge for cost of materials, most everyone will go out, pop in the new Kickstart, and be on their way. And with the overwhelming prevalence of 1.2 over 1.1 (hasn't happened yet, waiting on that FINAL release…), 1.1 becomes a thing of the past. If everyone had to open their machines and swap ROMs, it would take much longer for updates to be globally in place, and the percentage of holdouts could be large enough to form a critical mass. At this point, developers would write for both systems, avoiding any enhancements in the new system. -Hazy
#34536From: Don Curtis/SYSOPSep 27, 1986 9:32 PM
Bela, Aren't guessing games fun! <grin> I was speculating, and I do see one obvious problem….the daughter board takes up one of the ROM sockets (which socket I presume would be needed for the OS ROMs) so we are really left with the ST style OS, you can have it in RAM, but lose 256K or you can have it in ROM and have full memory. Even with the ST's OS (TOS) in ROM, you can boot up the TOS disk and load the OS from disk…but do lose the RAM. One other somewhat obvious solution is to not offer the ROM upgrade to A1000 owners, but keep that OS on disk and put the ROM's only in upgrade machines. I mean Hazy did say "Not what everyone thinks……….. Don
#34521From: Bela LubkinSep 27, 1986 7:26 PM
If they put it in ROM, they will surely allow you to boot KickStart off a floppy as well. That's just common sense. However, the idea of having a ROM as well as the KS RAM is new to me. 256K of RAM is not a big expense, but I'm not sure about the circuitry that allows it to be write-protected. (Seems pretty simple, but what do I know). So maybe they've gotten rid of the WCS daughterboard, and just added another 256K to the standard memory, with the write-protect hardware, and the ROM. Yes, then you could boot ROM to WCS RAM, and then as the disk-resident part loaded, it could patch the WCS copy of the ROM as necessary. A neat way to have patchable ROM. Or how about this: 256K of battery-backed RAM. No ROM. You have to use the KickStart disk the first time you boot the machine (unless it's done at the factory), but then it isn't needed again until you get a new version of KS. Now that would be >nice<… – Bela
#34522From: Bela LubkinSep 27, 1986 7:26 PM
It sounds like there may be an error or ambiguity in one of CIS' B-protocol documents, that causes people to write different programs that bomb in the same way. I know "B" does work correctly all the time with some programs that support it. I'll see if Chuck Forsberg knows anything about it… – Bela
#34545From: Richard WebbSep 27, 1986 10:29 PM
If you'll be talking to Chuck Forsberg, be sure to ask him when Pro-YAM will be out for the Amiga. 😉 P.S. I use /slant/ for italics as well — and more often than _underline_ or >boldface< since the / character is "lower case", saving wear and tear on the little finger.
#34545From: Richard WebbSep 27, 1986 10:29 PM
If you'll be talking to Chuck Forsberg, be sure to ask him when Pro-YAM will be out for the Amiga. 😉 P.S. I use /slant/ for italics as well — and more often than _underline_ or >boldface< since the / character is "lower case", saving wear and tear on the little finger.
#34589From: Tom CattrallSep 28, 1986 11:03 AM
The B protocol code I use (not on my Amiga) is adapted from a procedure I got from a Borland DL. It has worked flawlessly on both directions so I believe as you do that CIS and B protocol aren't to blame. However, besides converting Turbo Pascal to UCSD I made a lot of mods to the error recovery code which I found to be very lax. Whoever wrote that procedure didn't stop to look at his code when he finished to think through what would happen in various error situations. Sounds like maybe the Amiga implementors stopped too soon as well. I could upload the code if anyone is interested. It's a UCSD Pascal procedure to be called by a terminal program whenvever an enq is received (05). It handles a complete B protocol transfer and then returns to the caller.
#34589From: Tom CattrallSep 28, 1986 11:03 AM
The B protocol code I use (not on my Amiga) is adapted from a procedure I got from a Borland DL. It has worked flawlessly on both directions so I believe as you do that CIS and B protocol aren't to blame. However, besides converting Turbo Pascal to UCSD I made a lot of mods to the error recovery code which I found to be very lax. Whoever wrote that procedure didn't stop to look at his code when he finished to think through what would happen in various error situations. Sounds like maybe the Amiga implementors stopped too soon as well. I could upload the code if anyone is interested. It's a UCSD Pascal procedure to be called by a terminal program whenvever an enq is received (05). It handles a complete B protocol transfer and then returns to the caller.
#34589From: Tom CattrallSep 28, 1986 11:03 AM
Bela– The B protocol code I use (not on my Amiga) is adapted from a procedure I got from a Borland DL. It has worked flawlessly on both directions so I believe as you do that CIS and B protocol aren't to blame. However, besides converting Turbo Pascal to UCSD I made a lot of mods to the error recovery code which I found to be very lax. Whoever wrote that procedure didn't stop to look at his code when he finished to think through what would happen in various error situations. Sounds like maybe the Amiga implementors stopped too soon as well. I could upload the code if anyone is interested. It's a UCSD Pascal procedure to be called by a terminal program whenvever an enq is received (05). It handles a complete B protocol transfer and then returns to the caller. Tom
#34589From: Tom CattrallSep 28, 1986 11:03 AM
Bela– The B protocol code I use (not on my Amiga) is adapted from a procedure I got from a Borland DL. It has worked flawlessly on both directions so I believe as you do that CIS and B protocol aren't to blame. However, besides converting Turbo Pascal to UCSD I made a lot of mods to the error recovery code which I found to be very lax. Whoever wrote that procedure didn't stop to look at his code when he finished to think through what would happen in various error situations. Sounds like maybe the Amiga implementors stopped too soon as well. I could upload the code if anyone is interested. It's a UCSD Pascal procedure to be called by a terminal program whenvever an enq is received (05). It handles a complete B protocol transfer and then returns to the caller. Tom
#34522From: Bela LubkinSep 27, 1986 7:26 PM
It sounds like there may be an error or ambiguity in one of CIS' B-protocol documents, that causes people to write different programs that bomb in the same way. I know "B" does work correctly all the time with some programs that support it. I'll see if Chuck Forsberg knows anything about it… – Bela
#34522From: Bela LubkinSep 27, 1986 7:26 PM
It sounds like there may be an error or ambiguity in one of CIS' B-protocol documents, that causes people to write different programs that bomb in the same way. I know "B" does work correctly all the time with some programs that support it. I'll see if Chuck Forsberg knows anything about it… – Bela
#34522From: Bela LubkinSep 27, 1986 7:26 PM
It sounds like there may be an error or ambiguity in one of CIS' B-protocol documents, that causes people to write different programs that bomb in the same way. I know "B" does work correctly all the time with some programs that support it. I'll see if Chuck Forsberg knows anything about it… – Bela