[Dwarf-discuss] Re: optimized code debugging

Bishop, John E john.e.bishop
Thu May 26 20:55:55 PDT 2005


I agree: don't forbid them.

I don't believe I suggested that DWARF define
a semantics; I can imagine compilers which don't
have the problem cited (can't tell what the
addresses are going to be), and I can imagine
that those compilers might want to agree 
privately with their local debugger on another
interpretation.  That suggestion was my attempt
to motivate not forbidding.  It was not an attempt
to define a special meaning to zero-length
ranges.

I'm persuaded by Jim Blandy's "argument from compiler
ignorance" that not only should zero-length PC-ranges
be ok, but so should negative-length ones also be.
Further, DWARF should not give either zero or negative
-length PC-ranges any meaning.

Or is the extension to negative-lengths not needed?

	-John  

Do we have a way to spell "ignorable" in
spec-speak?





More information about the Dwarf-Discuss mailing list