|Generating code for debugger (-g option). email@example.com (1992-09-16)|
|Generating code for debugger (-g option) firstname.lastname@example.org (1992-09-17)|
|From:||email@example.com (Steve Simmons)|
|Organization:||Engineering, CONVEX Computer Corp., Richardson, Tx., USA|
|Date:||Thu, 17 Sep 1992 13:05:53 GMT|
> I wonder what are the issues involved when a compiler emits code to be
> used by a debugger. For instance, if the symbol table
> (compile-time-environment) is to be passed down, then is there an
> established way of representing the symbol table; as well as other
> debugger related info?
The code generation by the compiler is usually the same as without a "-g"
switch. The problems lie in generating the corresponding symbolic
information for tracing the generated code back to the original source
back to the original source.
SIGPLAN's PLDI '92 conference proceeding has a paper that I co-authored on
how Convex's debugger CXdb tracks optimizations.
Also, stabs and dwarf have extremely well defined interfaces for
non-optimized code. In fact, they support more languages with their
interface. Their symbol tables tend to be the union of features in all
Return to the
Search the comp.compilers archives again.