Laurynas Biveinis <laurynas.bivei...@gmail.com> writes: > [Third try. Apparently the compressed dump was still too big to get through] > > So I got fed up with trying to navigate gengtype maze of type_p, > pair_p and others and trying to figure out the difference between > GC_POINTED_TO and GC_USED and what is so "maybe" about > GC_MAYBE_POINTED_TO, thus I extended gengtype to dump its data > structures representing GCC data types and variables. > > I wanted to attach an example dump it produced but it is about 350K > compressed and so it's too big for this mailing list. It's not > perfect since I > did depth-first processing and the very first type goes very deep > indeed by pulling virtually all of the remaining types and taking > perhaps 80% of the output. However it is already much more usable and > greppable for looking into interesting types than trying to do the > same in GDB. > > I will commit this to gc-improv branch shortly. I think this is > interesting enough to have on the trunk too and I will submit together > with some other patches from gc-improv in couple of weeks. > > Any comments?
This looks like a good idea to me. The patch looks fine. Ian
