Leisner.Henr@xerox.com (Marty) (07/19/89)
I found the cross-reference listings valuable until I got my disks. Once I got the disks, it's easier to boot minix and grep then wade through the listings -- maybe it makes sense to forget the listings or put the listings in a separate volume (I guess something like Knuth's "Computers and Typography" set). In general, I like Doug Comer's style a little bit with the Xinu series -- I found it easier to relate the text to the source code. But I don't think he has nearly as much source code. Once you have the files on disk, there are lots of tools to interrogate the software. Is anyone using Greg McGary's dbcr package (a C symbol database)? Its been posted to usenet and is on uunet and simtel20. I've been using it on Ms/Dos and Unix for a long time and find it invaluable. You can feed it file names and it tells you distinct identifiers (or identifiers not distinct within n characters and other such stuff). You feed it a symbol and it tells you which files use the symbol. It builds a database on disk so its fast. It has front ends for running grep and editors by only invoking them on the files which contain the list identifier and automatically searching the selected identifier. It should compile on Minix (at least cross compile) -- I'm not sure if I ever got it running. The only problem is it seems to be core-bound -- i.e. I'm not sure it can do all of Minix on a 808x -- if the database were done on a bigger machine I'm not sure the access routines could easily access it. marty ARPA: leisner.henr@xerox.com GV: leisner.henr NS: leisner:wbst139:xerox UUCP: hplabs!arisia!leisner