home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: gnu.gdb.bug
- Path: sparky!uunet!cis.ohio-state.edu!ips.CS.tu-bs.DE!neitzel
- From: neitzel@ips.CS.tu-bs.DE (Martin Neitzel)
- Subject: Re: Execution trace feature for gdb
- Message-ID: <1992Jul28.114950.3647@ips.cs.tu-bs.de>
- Summary: It's not sooo easy...
- Sender: gnulists@ai.mit.edu
- Organization: Inst. f. Informatik, TU Braunschweig, FRG
- References: <9207201607.AA02430@bugs.airs.com>
- Distribution: gnu
- Date: Tue, 28 Jul 1992 11:49:50 GMT
- Approved: bug-gdb@prep.ai.mit.edu
- Lines: 21
-
- I just asked the same question about the "trace facility in the dbx
- sense" three months ago and digged into the gdb source.
-
- I found out that this functionality would require more than, say, a
- twenty line patch. The framework for integrating "trace points" as
- another variant of break and watchpoints points is there. The new
- "trace" commands set would nicely fit into it (arguments/syntax just
- like "break" commands with lots of code reuse possible). I'd estimate
- that modifications would mainly concern about three modules.
-
- But at that time, I was just too busy with my regular work(*) to
- figure out how to distribute the trace marks into the code (for a
- "trace function fred" command). I felt not being expert enough to
- assess viability/quality my own idea for an implementation and didn't
- any further work on it. (A broad, two page overview about gdb's inner
- concepts and workings would have been very helpful at that point.)
-
- Martin Neitzel
-
- (*) Still am, sorry.
-
-