hatcher@INGRES.BERKELEY.EDU.UUCP (06/13/87)
Since I haven't heard this mentioned before, I thought I'd pass on this bug: if there is a hard error on the diskette in the middle of an icon, you get a guru. I don't have the number handy but it's real reproducible....just open the window containing the bad icon, and when workbench tries to read the icon, ka-blam! Obviously workbench is not checking for errors, and it proceeds with bad information in the icon structure. Reading the same icon with a custom written C program just produces the read/write error requestor, as expected; no crash. I don't know if it matters, but the disk in question had a whole directory full of icons, and several tracks were bad, trashing several icons in the directory. My C program successfully examined all files on the disk without producing anything more than the read/write error requestor. Doug Merritt ucbvax!ingres!hatcher (thru Jun 28) or ucbvax!unisoft!certes!doug
paulz@hpspdla.UUCP (06/16/87)
Your program would appear to be a good tool for recovering from defective disks. Any chance of distributing it?
hatcher@INGRES.BERKELEY.EDU.UUCP (06/20/87)
In article <4160002@hpspdla.HP.COM> paulz@hpspdla.HP.COM (Paul Zander) writes: >Your program would appear to be a good tool for recovering from defective disks >Any chance of distributing it? Sorry, didn't mean to mislead anyone. I use "disksalv" to recover defective disks, and am not aware of anything better. The program I referred to was simply reading files for another purpose (my as-yet-unreleased "filetype" program), and I mentioned it simply to prove that simply reading the trashed icon files would not trash the system, but instead it was what Workbench did *after* reading the trashed file that crashed the system. Presumably it doesn't notice that the read of the file didn't work, and proceeds with the icon data structures that it *thought* it read, causing a crash. Doug Merritt ucbvax!ingres!hatcher (thru Jun 28) or ucbvax!unisoft!certes!doug