Re: debuggers - request for information

>[Creative new untraditional debugger approaches?]


The line between debugging and tracing is vague: the basic purpose of a
debugger is to take a wide range of state and map it to a narrower
range of state that a human can grok. Many tracing tools do much the
same thing.


For example, the EDSAC debugger I keep mentioning is also a program
tracing tool. (And, the Shade simulation and tracing tool I keep
making noise about has been used to implement a debugger.) I'd also
check out the program visulatization tools such as the UW Illustrating
Compiler (UW TR #90-07-01; see under `http://www.cs.washington.edu'),
though I don't have a good bibliography on such tools.


Blowing my own horn: for more about tracing tools (and some other tools)
see the related work section in the Shade SIGMETRICS '95 paper,


%A Bob Cmelik
%A David Keppel
%T Shade: A Fast Instruction-Set Simulator for Execution Profiling
%J Proceedings of the 1994 ACM SIGMETRICS Conference
on the Measurement and Modeling of Computer Systems
%D May 1994
%P 128-137


available under


http://www.cs.washington.edu/research/compiler/papers.d/shade.html
ftp://ftp.cs.washington.edu/pub/pardo/shade.ps.Z


Some of the tools that are oriented towards a user-level view of
program behavior are:


Dynascope [Sosic]
Parasight [Aral & Gertner]
Pixie
Purify


There are probably others; if you know of a tool that is missing from
the list, please let me know.


;-D on ( The bitwise paper chase ) Pardo
--

Return to the comp.compilers page.
Search the comp.compilers archives again.