Known bugs in ADB

June 20, 2000 -- when using test_adb with variable key sizes, keys
occasionally inserted in 'wrong' node. (looks like _another_ problem
with split)
  FIXED: June 20, 2000. Cached value of nth_keyLen(m) on
  stack. Compact()ion was apparently zeroing value.

June 21, 2000 -- allocation (bAlloc:alloc) doesn't check for overrun of freemap.
  FIXED: June 21, TODO: make this routine more efficient (store in tree?)

June 22, 2000 -- test3 produces truncated data (1K data, 1K nodes)
	june 22: appears to be related to independent (only?) nodes being flushed to disk, re-read
		data is corrupted (truncated to 1024 bytes suspiciously enough) when read from disk in nodeBuf insert node
	FIXED: June 21 -- allocated nodeSize/blockSize + 1 blocks instead of nodeSize/blockSize (/ is floor not ceiling)

June 23, 2000 -- occasional fatal error: attempt to flush node not in memory.

June 23, 2000 -- assertion "pktlen < pktsize" failed: file "/home/ny1/fdabek/src/sfs1/arpc/axprt_stream.C", line 300 triggered in sfsrocd.
    Incorrect hash values reported by sfsrocd (as well as correct
values) exist in database.