🚀 go-pugleaf

>Gavin.Flower@comp.vuw.ac.nz (Gavin Flower) wrote:
>
>In article <1992Nov2.152327.27296@cs.nott.ac.uk> smb@cs.nott.ac.uk (Simon
Burrows) writes:
>>That ADFSbuffers problem has been fixed between the two versions of RISC OS 3.
>>Unfortunately another one has reared its head in 3.10, but I believe it only
>>has an effect on 1Mb machines.
>
>What do you mean *only*, I have a 1Mb A310.  I can maybe afford to
>upgrade to Risc-OS 3.1, but not to more memory.  This serious, is it
>worth me upgrading at all?
>
>What is the bug, and is there a way round it - without having to
>disable the buffers?

Ah - an appropriate way to return to comp.sys.acorn.tech. Yet another
message on the perennial subject of ADFSBuffers. Yes, the 3.00 bug has been
fixed in 3.10. Yes, there is a new bug in 3.10. The 3.10 bug only occurs
when accessing floppy disks on machines with an 82710 or 82711 controller
ie. the A5000 or later. If you have an older model, it will not happen to
you. On A5000s and later, what usually happens when the bug strikes is a
total machine stiff-up while doing something to a floppy. So your hard disk
is safe. The thing about 1MB machines is bogus, and stems from the fact that
the fewer ADFSBuffers you've got, the more likely the bug is to happen.
Under 3.10, 1MB machines default to 2 buffers whereas 2MB uses 8 and 4MB
uses 16. So for a while it was thought that the bug was a 1MB bug. My advice
is to turn ADFSBuffers up to about 16 to make the chances less likely, and
just accept the occasional total stiff on floppy (it hasn't happened to me
yet and I'm using an A3020) unless you're using floppies a lot for critical
data. If you've got another copy of everything on the floppy, then there's
no need to worry. The speed increase of ADFSBuffers is well worth the chance
of the odd stiff, particularly when doing multitasking Filer_Action
operations. A fix for the bug is in the pipeline.

Owen.