Andreas Rossberg
Andreas Rossberg
The deeper reason I think was that we regard traps a different class of event. Unlike exceptions, traps are a fatal failure that should never happen in correct code. That...
What @tlively said. In addition, the occurrence of UB in a language like C/C++ actually means that all past(!) and future behaviour of the same program execution is undefined as...
@titzer, @dschuff, I seem to recall that we converged on throwing versions of the relevant instructions for that purpose during the f2f meeting. That's certainly cleaner and simpler than a...
Very cute! Question: I'm a bit confused about all the `/*empty*/` productions you inserted everywhere, and which seem to make the grammar incorrect. What are they for? Also, there appears...
First we have to write it. :) But I plan to be doing that once reftypes has been merged into the main spec.
Update: most of the spec has been there for a while, but still missing: * [x] Fix auto-publish. * [x] Track initialization status of locals, as recently decided. * [x]...
The spec is completed.
@conrad-watt, yes, @dhil will be on site, but @slindley can't come. I took the liberty to assume that @fgmccabe won't mind either way, but let us know if that's a...
@yamt: > if there are two failure modes, does it make sense for an engine to have an api to distinguish them? Probably. I guess it depends on what the...
@yamt: > i meant that this test case seems relying on a particular way to recover from a trap during instantiation. that is, do not free the instance. i guess...