dillon@CORY.BERKELEY.EDU.UUCP (10/28/87)
I've been getting my TEST: sample dos device driver working with the workbench.... I should have done this in the first place! the workbench is an excellent debugging tool for the driver! In anycase, I've found (and people have mailed me with) numerous bugs in the driver and a couple of undocumented 'features' that need to be implemented as well. I'll post an update after I get the sample driver working completely with the WB. Workbench bugs: (?) When the WB locks a file with intent to open "" (0-len str) it passes an illegal value for the Mode of the lock... neither ACCESS_READ or ACCESS_WRITE ... in fact, it seems to be another pointer to the lock! Strange... Workbench nitty gritty: Workbench relies heavily on proper timestamping. Proper timestamping for filesystems is not documented at all, so I will have to fool around with the floppies to figure it out. Improper (read none) timestamping screws up the workbench's caching scheme. Bugs found in TEST so far: -NOP's in assembly need to be before the label, not after -shared access to a file not implemented properly (multiple readers and writers). -directory scanning messes up under many multi-access conditions. -various other bugs -Matt