[LLVMdev] Requirements for the EH representation

[John McCall]

On Apr 13, 2011, at 2:06 AM, Renato Golin wrote:
> I'll read your email with greater care along the week, but I have
> added a page on the wiki to help us organise our thoughts.
> 
> http://wiki.llvm.org/Exception_Handling

If you want to volunteer to keep track of the discussion on the wiki, I
think that would be great.  I don't think we're going to hold the
discussion on the wiki, though.

Also, what you've listed as "Areas to Consider" are explicitly things
I *don't* want to consider immediately.  Those are interesting
questions, but they're separate questions, and we should settle
on these basic requirements first.

For example, if it's a real goal to make LLVM IR capable of supporting
a wider range of unwinders than it currently can — at a gloss,
non-resumptive EH with a single, non-reentrant landing pad per region
and LSDA layout approximating what's consumed by __gxx_personality_v0
— then that touches on almost everything else.  Contrariwise, if we're
only really interested in finding a better fix to the current IPO problems,
then maybe we can get away with very modest changes.  And it's okay
to have limited goals!   I personally don't;  I think we should aim to get
the IR design good enough to support crazy resumptive languages
with crazy custom unwinding schemes.  But I need to know what range
of problems we're willing to consider solving before I can usefully weigh
different solutions.

John.