mroussel@alchemy.chem.utoronto.ca (Marc Roussel) (03/30/91)
In article <1991Mar29.194707.15835@evax.arl.utexas.edu> cs4344af@evax.arl.utexas.edu (Fuzzy Fox) writes: >GEOS files are supposed to be type USR, because they are formatted in a >way that is rather alien to Commodore DOS. This is why you should never >validate a GEOS disk unless you are in GEOS. <<Buzzzz>> Wrong! GEOS files are indeed USR files, but that's not the part that confuses Commodore DOS. (USR files are just glorified SEQ files. Why GEOS chose to use the USR designation, rather than SEQ or PRG, we can only speculate.) The reason you should never validate a GEOS disk outside of GEOS is that Commodore DOS doesn't know about the border blocks. GEOS needs a way to keep track of files on the desktop's border. When you drag a file to the border, GEOS keeps track of this by creating an entry in a dummy directory in some specially designated blocks. (At that point, the files have been taken out of the regular directory.) To keep from overwriting this border directory, GEOS allocates these blocks in the BAM at format time. There are however no actual files kept there, so when you use the regular Commodore validate command on GEOS disks, it sees empty blocks and deallocates them in the BAM. GEOS' validate on the other hand knows about the special meaning of these blocks and leaves them alone. Marc R. Roussel mroussel@alchemy.chem.utoronto.ca
cs4344af@evax.arl.utexas.edu (Fuzzy Fox) (03/30/91)
In article <1991Mar30.021857.22494@alchemy.chem.utoronto.ca> mroussel@alchemy.chem.utoronto.ca (Marc Roussel) writes: > > <<Buzzzz>> Wrong! When people who use this game-show oriented phrase it really pisses me off, so forgive me if I flame you back. > GEOS files are indeed USR files, but that's not the part that >confuses Commodore DOS. (USR files are just glorified SEQ files. Why >GEOS chose to use the USR designation, rather than SEQ or PRG, we can >only speculate.) This is not right. It is true that Commodore DOS treats USR files like SEQ, but GEOS has invented a new filetype which is nothing like REL or SEQ, and it is called VLIR. Many data files, such as GeoWrite or GeoPaint, are in this format. A normal SEQ file has pointers to just the first data block, and that block points to the next, and so on. DOS treats USR files like this, too. A VLIR file from GEOS has 2 pointers, one to the header block and icon block, and another pointer to the record block. The record block, in turn, has pointers to 127 other records. Essentially, this block is a list of track/sector pointers, and DOS does not understand that at all, so if you validate with DOS, all of the actual data stored in a VLIR file will be de-allocated, resulting in tragedy later on. Furthermore, under GEOS, VLIR or SEQ files both have an icon block stored with the file, and DOS will free that up under validate, too. Next time you write to the disk, LOTS of data will get written over, and you probably won't even be able to open the disk anymore. >The reason you should never validate a GEOS disk >outside of GEOS is that Commodore DOS doesn't know about the border blocks. >GEOS needs a way to keep track of files on the desktop's border. This is also true, but it is not even close to the whole story. Of course, even if this were the only problem, a trashed out border block would still render the disk unopenable, and thus useless under GEOS. If you accidentally validate a GEOS disk, immediately go into GEOS and validate it there, to prevent losing any data. -- David DeSimone, aka "Fuzzy Fox" on some networks. /!/! INET: an207@cleveland.freenet.edu / .. Q-Link: Fuzzy Fox / --* Quote: "Foxes are people too! And vice versa." / ---
mroussel@alchemy.chem.utoronto.ca (Marc Roussel) (03/30/91)
In article <1991Mar30.044950.3816@evax.arl.utexas.edu> cs4344af@evax.arl.utexas.edu (Fuzzy Fox) writes: >In article <1991Mar30.021857.22494@alchemy.chem.utoronto.ca> >mroussel@alchemy.chem.utoronto.ca (Marc Roussel) writes: >> [Tiny small amounts of partly correct information.] [Much perfectly correct information about GEOS files.] Fuzzy Fox is right, of course. I should never post anything late at night. I'm sorry for the confusion I caused. Now I'll just go away and write a hundred times "I won't post answers to the net without researching them first." Marc R. Roussel mroussel@alchemy.chem.utoronto.ca