This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [RFA] Integrated hashtables for compiler + cpplib
- To: zackw at Stanford dot EDU
- Subject: Re: [RFA] Integrated hashtables for compiler + cpplib
- From: Mark Mitchell <mark at codesourcery dot com>
- Date: Sat, 19 May 2001 14:26:49 -0700
- Cc: neil at daikokuya dot demon dot co dot uk, gcc-patches at gcc dot gnu dot org
- Organization: CodeSourcery, LLC
- References: <20010519121704.L774@stanford.edu>
>>>>> "Zack" == Zack Weinberg <zackw@Stanford.EDU> writes:
Zack> and on a casual sweep I don't think any of them _need_ the
Zack> return-0- if-not-found behavior. Some of them look flat out
Zack> wrong to me.
I agree.
For example, the assemble_name one is probably bogus. It is supposed
to be an invariant that TREE_SYMBOL_REFERENCED is set for things that
are output. So, if there's no `id' for this thing, that indicates
that either:
- We are trying to mark the wrong thing. Some amount of hokey
back-end mucking with identifiers has confused us.
- We should create one, and mark that, in case anyone else
talks about in the future.
I don't know which is right -- but I think the current behavior is
probably wrong.
The only thing we want to be careful is that we don't somehow create a
bazillion extra identifiers for `_foo' because that's how this target
spells `foo' or something like that.
--
Mark Mitchell mark@codesourcery.com
CodeSourcery, LLC http://www.codesourcery.com