https://gcc.gnu.org/bugzilla/show_bug.cgi?id=119241
--- Comment #16 from Richard Biener <rguenth at gcc dot gnu.org> --- (In reply to Robert Dubner from comment #14) > "Wait for me! I am your leader!" > > I am not *used* to this. > > What I *am* used to is people telling me there's a problem in my code, and > then standing around tapping their feet impatiently until I fix it. > > Okay. Deep breath. Because I am totally curious, I am going to download > Richard's patch and apply it so that I can at least see what's going on. Note my patch is mostly an attempt to see what needs fixing when we move that 'value' thing to be a tree. It's for sure not correct as-is as I didn't understand most of the rationale behind the string processing and how important the exact format is (real.cc is limited as to both what radixes it can convert from string and to the output formats it supports to string). One of the advantages of using 'tree' going forward would be that we can store native integers therein as well, so we don't have to use floats for everything (some uses appear to make sure the floats are actual integers). But that's for a secondary change. > It has taken me a ridiculously long time to figure out that you are creating > trees for numeric values, and then doing arithmetic on those trees at > compile time. > > That's a methodology that simply didn't occur to me *at all*, and it's > causing the ground to shift beneath my feet. It's disconcerting. > > And what I am seeing here is a rampant enthusiasm bordering on glee for > improving the front end. And I simply have to get past the idea that *I* > have to fix everything. > > So: I am basically asking: "What can I do?" > > I offer up as an answer: "Bob, we need more test cases." Definitely yes! Also eventually explaining what some of the code actually wants to do. I'll continue poking around until I get a least hello world working (need to fix^work around that global initializers tomorrow), and then will think of taking the whole thing apart into easier to review and test pieces. > Does that sound right?
