I was about to answer on that point. This fix was made to fix a bug on top of a modified MAsm stressing literal pools generation. I don't see how to create a good and predictable test with the current code base. Manually generating the bug would require fine control over literal pools and debug slots generation. For now we don't have such deep control over code generation from the tests.
So I agree we need one, but I think it should come with the incoming work on strengthening this mechanisms (See Rodolph's comment on the other related patch), which should provide options to influence and test literal pools. Alexandre Regarding Michael's comment, I think such a test is hard to create. Manually creating a situation where the bug occurs requires fine control over the literal pools and debug slots generation. Tests don't have such deep control over code generation. I think we should try to include a test when we introduce additional checking and control mechanisms around the literal pools generation. On Thu, Jun 14, 2012 at 12:07 PM, <[email protected]> wrote: > LGTM. Landing. > > Though Michael has a point. We would certainly welcome a regression test > for > this. > > https://chromiumcodereview.**appspot.com/10449047/<https://chromiumcodereview.appspot.com/10449047/> > > > -- > v8-dev mailing list > [email protected] > http://groups.google.com/**group/v8-dev<http://groups.google.com/group/v8-dev> > -- v8-dev mailing list [email protected] http://groups.google.com/group/v8-dev
