On Fri, Nov 07, 2014 at 08:19:26PM +0100, Petr Machata wrote: > Mark Wielaard <m...@redhat.com> writes: > > > If it isn't reserved then we should indeed add a dwarf_macro_version () > > function. But I hope we can just keep one function to iterate the > > macros. Can we use "1" as special token, instead of "0" for > > dwarf_getmacros () to start "new style capable users" iteration? Then we > > have a way to know whether the user "promises" to use > > dwarf_macro_getparamcnt and dwarf_macro_param instead of old style > > accessors. > > 1 is a valid offset in .debug_macinfo. I think that a named constant > that evaluates to PTRDIFF_MAX would work. Together with a version-query > interface, this would work well for supporting the dual nature of 0xff.
Ah, yes, that would work well. > Do you want the change in right away? I'd wait for the committee > decision before making it. Otherwise we should add dwarf_macro_version > as well. Lets get your patch in first. I liked the changes you made. Please squash those follow up commits before pushing to master. We shouldn't do a release with this new interface before hearing back from the DWARF committee though. So maybe just add a "FIXME: 0xFF opcode handling!" note to the NEWS entry, so we don't forget. Cheers, Mark
