>>>>> "Dodji" == Dodji Seketeli <do...@redhat.com> writes:
Tom> One thing to note is that there may be some code that assumes an Tom> ordering of location values. If you hand out ordinary and macro Tom> locations separately (which I think is what is going on), then code Tom> doing this may break. [...] Dodji> Now I think I am left with two options, I guess. Either I find a way to Dodji> make the two kind of locations (those mapped into macro maps and those Dodji> mapped into ordinary maps) share the same integer space and be ordered, Dodji> or, I spot the places in the code that assume an ordering of all Dodji> location values and I change that. Dodji> Is the location ordering a strong property we want to keep? It doesn't matter to me. I think it would probably be sufficient if there were a "cheap-enough" comparison function. Dodji> I haven't noticed that with my tests. But after reading this, I have Dodji> found one spot of the in diagnostic_report_diagnostic that does this. I Dodji> don't know yet if there are other places. Too bad we can't just write operator< :-) Dodji> But if you want, I can just move the the comments to the header file, Dodji> no problem at all. Yeah, please do. Tom
