On 16 March 2015 at 17:52, David Malcolm <dmalc...@redhat.com> wrote: > I've been experimenting with revamping our diagnostics to better show > source locations. Some of the ideas are inspired by Clang's > diagnostics, but I'm going beyond what it does in some areas. > > I'm attaching a patch (against r221423), which is very much a > work-in-progress, but in a "release early, release often" spirit I > thought I'd solicit feedback here (obviously this would be for our next > stage 1/gcc 6):
Wow, awesome! Obviously, I'm very much in favour of this in general. >From my own experience, you are going to have two enormous bike-shedding "experiences": one for the API and another for the output. My advice would be to implement a few obvious testcases and just the ability to print some more detailed location info, like Clang's -fdiagnostics-print-source-range-info (but see also my past troubles with that: https://gcc.gnu.org/ml/gcc-patches/2009-08/msg00174.html) without even printing any caret info. This way you can work out the internal details without having to defend at the same time the actual user interface. I think my mistake was to try to implement -fdiagnostics-print-source-range-info as a user option, I should have proposed it as a debug option such as -fdump-location-info and just move forward from that. Also, people are going to more easily review and accept this if you propose small incremental steps that just tackle (initially) one case at a time (of course, you may be working on your grand plan behind the scenes). Cheers, Manuel.
