revised PATCH: CALL_EXPR representation changes, part 0/9 (table of contents)

[Brooks Moses]

Jan Hubicka wrote:
> Hi,
> this is report from memory tester, overall it seems to be slight memory
> consumption increase. Are there any plans to get the memory back?

I'm not completely sure where the increase is coming from; Sandra may 
have a better idea.  My guess, though, is that this is coming from some 
of the places where the original TREE_LIST representation of the 
argument list hasn't been completely removed, and so the compiler ends 
up doing most of the processing with the new representation and then has 
to cons up a TREE_LIST version as well.  If that's what the increased 
memory consumption is coming from, there's definitely an intention to 
address that.  See the "remaining CALL_EXPR related cleanups" thread for 
details.

(And it looks like, while I was writing this, Paolo suggested the same 
thing.)

Thanks for checking this!

- Brooks