[Bug middle-end/30364] [4.1/4.2/4.3 Regression] Wrong variable ranges due to constant folding

[pinskia at gcc dot gnu dot org]

------- Comment #2 from pinskia at gcc dot gnu dot org  2007-01-04 10:23 -------
(In reply to comment #1)
> Is it really a middle-end issue?
Yes because fold is part of the middle-end that is most likely causing this
being translated into "a + b - 40".

And this is the reason why the way wrote it the second way works is because
fold does not see the full expression so it does not do the folding.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30364