[comp.unix.wizards] Does the 'dbx' debugger work?

drw@cullvax.UUCP (01/23/87)

I've been using the dbx debugger, and it seems to have some problems.
In particular, (1) sometimes when an identifier is subscripted, the
result doesn't seem to be the correct address, (2) the "-" operator
applied to two pointers doesn't always produce the difference,
measured in units of the objects pointed to, but rather the difference
of the raw pointers, and (3) sometimes when 'delete' is used to remove
a trace, that trace is later triggered by the code, and dbx crashes.
Also, putting in stops or traces seems to slow execution by much more
than I expect.

P.S. We're running Ultrix 1.2

Dale
-- 
Dale Worley		Cullinet Software
UUCP: ...!seismo!harvard!mit-eddie!cullvax!drw
ARPA: cullvax!drw@eddie.mit.edu

chris@mimsy.UUCP (01/24/87)

I cannot say whether the Ultrix 1.2 dbx is broken.  But re:

In article <735@cullvax.UUCP> drw@cullvax.UUCP writes:
>Also, putting in stops or traces seems to slow execution by much more
>than I expect.

How much do you expect?

Adding `stop at' is easy.  Adding `stop when' or `trace' is hard.
`Stop at' is easy because a line translates to a particular machine
instruction address, and dbx can insert a breakpoint there.  `Stop
when' or `trace' is hard because the only way dbx can do this is
to single step every instruction, and check after each instruction
to see whether an appropriate change has occurred.

Basically, dbx needs better control over processes than is provided
by present kernels.
-- 
In-Real-Life: Chris Torek, Univ of MD Comp Sci Dept (+1 301 454 7690)
UUCP:	seismo!mimsy!chris	ARPA/CSNet:	chris@mimsy.umd.edu