home *** CD-ROM | disk | FTP | other *** search
- Path: sdcc12.ucsd.edu!sdcc13!ckaiser
- From: ckaiser@sdcc13.ucsd.edu (Po-Ching Lives!)
- Newsgroups: comp.sys.cbm
- Subject: Save-replace bug (was Re: Various Commodore Computers (please read))
- Date: 25 Jan 1996 16:00:24 GMT
- Organization: University of California, San Diego
- Message-ID: <4e89eo$1dg@sdcc12.ucsd.edu>
- References: <577469@363.chatlink.com>
- NNTP-Posting-Host: sdcc13.ucsd.edu
-
- In <577469@363.chatlink.com> Bios@sys363.chatlink.com writes:
-
- [ NA> = included from a previous article ]
-
- >NA>As I recall (which is probably not entirely correct) is something with
- >NA>un-ready drives (such as drive #1 which isn't there) and a shortage of
- >NA>buffers (another 1540 problem, since the single drives are half a
- >NA>double drive in every respect: half the mechanics, half the processors
- >NA>and half the memory).
-
- A popular way of putting this is that the 15xx drives, and I think
- early 1571's, with the exception of the 1581, spend most of their
- time convincing themselves they're really drive 0, and not drive 1.
- So, you get a ghost phenomenon where in certain isolated
- circumstances, you have a drive with multiple personalities that
- writes to this ghost drive 1 (the theory goes) and makes a mess.
-
- >NA>There was apparently a program published that reliably demonstrated the
- >NA>bug (even though it took a while). I'd like to see that program and
- >NA>the related article (Transactor?).
- > Yes, I have an article in my transactor magazine which actually
- > demonstrates this bug. Never tried the program though but it seems
- > to prove that there is a bug with the save and replace feature.
-
- The save and replace feature's problem is that it attempts to keep
- two copies of the file on the disk at the same time. Normally, this
- is not a problem and is in fact a GOOD thing because if something
- happens to one copy, or the other, there's still an image of one of
- the files on the disk. However, should the ghosting problem happen
- to occur while one of these copies is being dealt with, hell breaks
- loose.
-
- I have in fact encountered the @0: bug many times myself and
- it has wrecked untold hours of work, which is why I now scratch and
- save instead. It is *supposed* to go away if you specify the drive
- number (save"@0:file",8 instead of save"@:file",8) but I have not
- found any increase in reliability with this method. Where there's
- smoke, there's usually fire. Anyone know if the emulators also have
- this problem?
-
- Cameron Kaiser
- ckaiser@ucsd.edu
- visit the CWI home page at http://www.armory.com/~spectre/cwi.html
-