LGTM too. The comment is not quite right and it confuses me just a little bit more than not having it :).
I think LoadCondition might emit *conditional* jumps to one or both targets as well, and still not leave cc_reg_ set. This would happen for instance with (exp0 && exp1) where exp0 has control flow to both targets and exp1 folds to an unconditional jump. has_cc() signals to the caller that LoadCondition did not emit the last test and branch it needed. You might phrase the comment positively in those terms: "LoadCondition may (and usually does) leave a test and branch to be emitted by the caller. In that case, negate the condition." On Wed, Jul 8, 2009 at 11:55 PM, <[email protected]> wrote: > Please add comment as discussed, otherwise LGTM. > -Ivan > > > > http://codereview.chromium.org/149351 > --~--~---------~--~----~------------~-------~--~----~ v8-dev mailing list [email protected] http://groups.google.com/group/v8-dev -~----------~----~----~----~------~----~------~--~---
