[tree-ssa] Splitting abnormal edges

[Andrew MacLeod]

On Tue, 2003-06-17 at 12:23, Jeff Sturm wrote:
> On 16 Jun 2003, Andrew MacLeod wrote:
> > 	* tree-cfg.c (find_insert_location): Append after last stmt in SRC when
> > 	block ends in a control altering stmt.
> 
> Looks like you've applied this patch.  It aborts building libjava, as it
> should... it can't insert on the bogus EH edges.
> 
> However, I'm trying to get the Java gimplify patch in shape to
> commit this week, and this sets me back a bit :(
> 
> It'd be nice to have EH completely fixed, but that's not easy, and Jeff
> seems to be working on it anyway.
> 
> I'd like to get the Java patch in so that people other than Andrew Haley
> and myself are testing it.  What can we do?  Would it be OK to simply
> revert this portion of the patch for now?
> 

Just remove the abort, or change the if in the default case to if (1) or
something...  You don't need a reverted patch in the tree to keep
working on your patch do you?

Once you are ready to commit your patch, then if EH isn't working we can
do something about it. I beleive we are trying to fix that right now...

Is that reasonable, or does it cause undue headache? 

Andrew