On Apr 19, 2013, at 2:27 PM, Marcus Denker wrote: > > On Apr 19, 2013, at 1:56 PM, Marcus Denker <marcus.den...@inria.fr> wrote: > >> >> On Apr 19, 2013, at 1:35 PM, Henrik Johansen <henrik.s.johan...@veloxit.no> >> wrote: >>>>> >>>> >>>> Did you use >>>> >>>> https://ci.inria.fr/rmod/job/Opal/ >>>> >>>> The Opal in the image is a bit out-dated… >>> >>> I did just now, but there, debugging *anything* (even 3 + 4, or using cmd >>> dot) with the Opal Compiler set to default seemed to bring the image into a >>> near catatonic state… >>> Sort of hard to check if the buttons were fixed, with no debugger ever >>> opening. ;) >>> >> >> Yes, we committed some changes that might have broken things… >> I will check. >> > Ok, found it… at least the debugger appears again after a "self halt" :-) > > https://ci.inria.fr/rmod/job/Opal/lastSuccessfulBuild/artifact/Opal.zip > > We should remember to never think it works before the regression tester is > successful… > Clement wrote a blog post: > http://clementbera.wordpress.com/2013/03/22/toward-a-zero-bugs-interpreter-compiler/ > > Marcus > > That's *really* great! With everything properly initialized, it works smoothly! (other than having to click more than once to advance PC into the optimized to:do: blocks)
One thing I ran into that doesn't work as I'd expect now, is UHE's when using "Through"; Debugging "nil asInteger" or, "ClosureTests new testToDoOutsideTemp" (which has collection uninitialized, and exhibits the same at #add: send) - pressing "Over" I get the expected behavior of opening MessageNotUnderstood: in the same Debugger - pressing "Through", I get the MNU in new pre-debugger after a loooong delay on first click, then, presumably with PC erroneously incremented, an unrelated error on 2nd click, also in new pre-debugger. Cheers, Henry
