I don't expect this to be a noticeable performance hit, since the force_allocation is a fully inlineable reference to a static variable. Since it's a correctness issue (causing crashes on Android) we have to commit it anyway, though we can certainly try to find a faster solution if it turns out to be noticeable.
http://codereview.chromium.org/251072/diff/1/2 File src/api.cc (right): http://codereview.chromium.org/251072/diff/1/2#newcode110 Line 110: printf("%s: %s\n", location, message); On 2009/10/02 12:53:33, antonm wrote: > is it a reminder of debugging or intended change? I guess it's a remainder of debugging. It certainly would have saved me a lot of time if it had been in there when I started investigating this issue. For some reason the VM crashes shortly after going though this function, so the message is lost. I'm going to change it to a PrintF that is only done in DEBUG mode. This routine is only called if no other error handler has been registered, so that seems harmless and potentially very useful. http://codereview.chromium.org/251072 --~--~---------~--~----~------------~-------~--~----~ v8-dev mailing list [email protected] http://groups.google.com/group/v8-dev -~----------~----~----~----~------~----~------~--~---
