|Books on Debuggers email@example.com (2004-04-28)|
|Re: Books on Debuggers firstname.lastname@example.org (2004-04-29)|
|Re: Books on Debuggers email@example.com (Paul F. Dietz) (2004-04-29)|
|Re: Books on Debuggers firstname.lastname@example.org (Joe) (2004-04-29)|
|Re: Books on Debuggers email@example.com (glen herrmannsfeldt) (2004-05-02)|
|Re: Books on Debuggers firstname.lastname@example.org (2004-05-02)|
|Re: Books on Debuggers email@example.com (2004-05-08)|
|Re: Books on Debuggers firstname.lastname@example.org (Tim Bauer) (2004-05-08)|
|[1 later articles]|
|From:||email@example.com (Nick Maclaren)|
|Date:||29 Apr 2004 12:05:06 -0400|
|Organization:||University of Cambridge, England|
|Posted-Date:||29 Apr 2004 12:05:06 EDT|
Mohd Hanafiah Abdullah <firstname.lastname@example.org> wrote:
>Could anyone please recommend any good books on how to write a debugger?
>[I've never seen one. -John]
I have never heard of even a competent one. There have been books on
how to write the user interfaces and high-level facilities, but most
of the aspects are the sort of messy engineering that was learnt by
apprenticeship and not written up. That is one reason that few modern
debuggers approach the usability of the best 1960s and 1970s ones.
I haven't written one, but I have written thorough and flexible error
handling in run-time systems (a related problem), and 90% of the
problems you hit aren't documented even in the operating system
specification. Sometimes not even in the HARDWARE specification :-(
Note that a debugger that can handle value errors in a more-or-less
corect program doesn't hit these, which is why so many demonstrations
of debuggers work. It is when the program has just trashed its stack
and executed a system call with random arguments that you hit these
Could such a book be written? I doubt it would be readable.
Return to the
Search the comp.compilers archives again.