Re: debuggers - request for information

Arne Watnelie (arnew@ifi.uio.no) wrote:
> [snip]
> 3) Have anybody made any creative, new debuggers who breaks the tradition
> of using breakpoints, stepping etc.? Implementing new, untraditional
> approaches to the debugging task?
> (Like when people suddenly made cursor-oriented editors instead of
> line-oriented/command-oriented editors.)
> [snip]


The SML-NJ debugger has a weird feature which lets you jump backwards
or forwards in time and view the program state.


This works by taking occasional snapshots of the system state. When
you jump back in time, you restore to the closest state and run
forwards to the time you wanted.


The snapshots are reported to be pretty small because SML programs are
mostly declarative and so contain few variable updates. I can't
remember how they deal with IO operations.


Internally, the debugger works by performing a source-to-source
transformation of the syntax tree adding any debugging info thatmay be
required. Sadly, this transformational approach imposes a high
overhead on execution time: a factor of 2 or 3.


%A AP Tolmach
%A AW Appel
%T Debugging Standard ML without reverse engineering
%B Proc ACM Conference on Lisp and Functional Programming, Nice
%I ACM
%D June 1990


Alastair Reid
Yale Haskell Project
--

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