1) Make it an editor rather than a displayer:
   a) move around in the block display menu and edit in hex or ASCII
   b) edit inode, block pointers -- for real disk repair

2) Clean up source:

   command line parser isn't too hot.
 
   Figure out why those writes were allowed
   in write_table() and map_block() -- if you
   run fsck with the repair flag off, it should
   not write anything to the disk on the off chance
   that it might hit something important.
   Right now they are changed so that they occur
   automatically, only if one of the repair flags
   is on (i.e. it is not interactive).

3) Implement search and recover function.  This 
   might not be that hard.  It will be almost impossible
   to reconstruct a file if an indirect block gets hit,
   but it should be easy to write something which will
   search for indirect block patterns (i.e. low/high byte
   pairs that are close to each other) and point out 
   unused blocks on the disk which are indirect or 
   2x indirect, then let the user rebuild the inode entry.
   
   Also, it is pretty easy to search the start of a block 
   for some kind of file magic, pick some useful things like
   gzip, ZIP, shell scripts, maybe even parse /etc/magic if
   it is there.  Perhaps we could just tear through the inode
   table and send the first block of each unused inode 
   through `file`. Or, we could do this for all the unused
   blocks, it may turn up a lot of bogus results, but it
   sure beats figuring things out by hand.

   (There is some preliminary stuff search code (if anyone
   can figure out why the spacing is so screwy, let me
   know), both pattern and indirect block, which will be
   included if EMERGENCY is defined.)

4) Implement block/inode select from block mode using the 
   cursor.  If the block is a directory, positioning the cursor on 
   an entry should allow the user to switch to inode mode viewing
   that entry.  Under block mode, the user should be have random 
   access to the indirect block using the cursor to select
   blocks of interest.

5) What about other file systems other than MINIX?
   xiafs shouldn't be too hard.

