Patch: darwin specific

Dale Johannesen dalej@apple.com
Mon Nov 25 11:07:00 GMT 2002


On Monday, November 25, 2002, at 10:38  AM, Zack Weinberg wrote:

> Dale Johannesen <dalej@apple.com> writes:
>
>> This uses the TREE_TYPE and TREE_CHAIN field of some IDENTIFIER
>> nodes, which are currently idle.  Per Bothner is working on
>> IDENTIFIERs and knows about this new usage.
>
> This interferes with my plans to make IDENTIFIERs not be trees.  Can
> you find an alternate implementation please?

I should explain this more.  There is a need to keep pointers, 
potentially
two, from one IDENTIFIER node to another; I discussed this with Stan 
and Mike
during development and got agreement that this was the best approach.
The ID's don't have to be trees as long as there's a place for the 
pointers,
and in fact the pointer space is not needed in most IDs and can be 
lazily
allocated, although it isn't now.  I believe Per is thinking along the 
lines
of having a target-specific area pointed to from the ID nodes.  The 
darwin.c
changes were made with the anticipation that the IDENTIFIER structure 
might
be changing, and I tried to set it up so it would be easy to maintain in
this case; see the DARWIN_ macros at the beginning of the patch.





More information about the Gcc-patches mailing list