This message is from the T13 list server.
> As for the email that started this discussion, it described a very > broken device. Eventually we fell into discussing the very ordinary case of bits of the upper byte stuck. Myself I'm asking to solve broken PATA cables, not broken device firmware. Rumour tells me 80 wire cables in particular break often enough for standard procedure in some development labs to be substitute a new cable before reporting an issue. That workaround isn't so very practical, for normal retail customers of nth-tier OEM's. > my software does not accept invalid values > in any ID word that is used I think with study we could prove a test like that does or does not detect all stuck faults in the hi byte. I wonder if that incorporating a test like that into commercial software could reject devices that otherwise work fine except by way of non-standard op xEC/A1 data. An interoperability result like that would then not encourage the adoption of the checksum. > there is no such thing as MWDMA mode 3, etc. That kind of test sounds less likely to break interoperability than others. Pat LaVarre
