This is the mail archive of the mailing list for the GCC project.

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [PATCH] Remove prototypes for long removed vn related functions

On Fri, Jun 05, 2009 at 09:44:29PM +0200, Richard Guenther wrote:
>On Fri, Jun 5, 2009 at 9:27 PM, Bernhard
>Reutner-Fischer<> wrote:
>> :ADDPATCH gcc:
>> ? ? ? ?* tree-flow.h (make_value_handle, set_value_handle, sort_vuses,
>> ? ? ? ?sort_vuses_heap, vn_lookup_or_add, vn_lookup_or_add_with_stmt,
>> ? ? ? ?vn_lookup_or_add_with_vuses, vn_add, vn_add_with_vuses,
>> ? ? ? ?vn_lookup_with_stmt, vn_lookup, vn_lookup_with_vuses): Remove
>> ? ? ? ?prototypes for removed functions.
>> ? ? ? ?(expressions_equal_p): Move to ...
>> ? ? ? ?* tree-ssa-sccvn.h: ... here and ...
>> ? ? ? ?* matrix-reorg.c: ... adjust includes.
>Huh.  expressions_equal_p should likely be private to tree-ssa-sccvn.c.  The

indeed. Freaky, isn't it. One would assume that there is existing infrastructure
for checking two tree's equality from some generic spot. But i can only find
rtx_equal_p and cp_tree_equal (without '_p'). Should, perhaps, both of sccvn
and matrix-reorg use some existing other tree-equality check?
Just curious..

I'm aware that gcc keeps growing and growing which provokes certain kind of
folks to curse periodically ;) Reusing common infrastructure, with all it's
pros and cons, may not be all that relevant, but IMHO has it's benefits.

>call in matrix-reorg.c is confusing...
>Well.  The patch is ok.

Thanks, will install for now.

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]