This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: [PATCH] Fixes tree-loop-distribute-patterns issues
- From: Roland McGrath <roland at hack dot frob dot com>
- To: Adhemerval Zanella <azanella at linux dot vnet dot ibm dot com>
- Cc: "GNU C. Library" <libc-alpha at sourceware dot org>
- Date: Tue, 18 Jun 2013 13:56:08 -0700 (PDT)
- Subject: Re: [PATCH] Fixes tree-loop-distribute-patterns issues
- References: <51C0AFB7 dot 1060009 at linux dot vnet dot ibm dot com>
What was discussed before was using a configure check. In general it's
better to use empirical checks rather than version checks, even if in
practice a version check is going to suffice. In this case, it does not
seem difficult to write a configure check.
Other macros of this sort we have are defined in lowercase
(internal_function, etc.). Some of those have names indicative of their
important semantics rather than echoing the syntax of the implementation.
I think we should use such a name for this case too. I haven't thought of
a particularly good name, but I don't think it should include "tree",
"distribute", or "patterns".
Attributes can appear in function definitions, right before the return
type. There's no need for a separate forward declaration.
The most obvious place we need these annotations is on the simple_*
functions in string/test-*.c, so I would start by adding it to all those.
That seems like the right initial set that goes in the change that
introduces the macro and serves as implicit documentation of its use.
Then we can take on each actual implementation that might be affected.
Thanks,
Roland