[ This note was written by Andrew Molitor. ] There's a binary here, in rawreport.c which walks your dbm database and runs out a raw report of the form: Object 214 resides at offset 31488 and takes 312 bytes Object 17 resides at offset 4352 and takes 7125 bytes etc.. It assumes you're running with the database patches applyed after 2.2.0. If you're not, look for the definition of struct hrec. right under the comment that says 'This struct should match the one in udb_ochunk.c' and delete the line about blox, the last line of the structure definition. Compile with 'cc rawdump.o -I.. -I../src -o rawdump' using cc or gcc, whatever your poison is. There's also a shell script, report.sh and an awk script useful for generating histograms from input. The basic plan is that you find your dbm database, for example if your database lives in files: foo.gdbm.db foo.gdbm.pag foo.gdbm.dir then your database is named 'foo.gdbm', it's the name before the '.pag' and the '.dir'. Then you can 'report.sh foo.gdbm' to get pretty pictures of stuff. Since the shell script wants to find both the raw reporter 'rawreport' and the awk script 'histogrammer' in the path, it's probably best to run this in the tools subdirectory, and point it at the database file with an explicit path. For example: report.sh ../../game/tinymush.gdbm or something. If your machine doesn't have 'nawk', change the definition of AWK in report.sh to be 'awk' instead of 'nawk', and hope. The histogrammer requires the newer version of awk, mostly called awk, but called nawk on Suns.