Re: [PATCH] Avoid T.NNN names for function clones

[Jakub Jelinek <jakub at redhat dot com>]

On Wed, Mar 25, 2009 at 11:03:41AM -0700, Roland McGrath wrote:
> Personally I would prefer that the original identifier be more clearly set
> off from the suffix than "_clone".  (The goal here is to be less confused
> reading a backtrace that has only symbol names and no DWARF info.) i.e.
> "foobar_clone.123" sends me looking for "foobar_clone" in the program.
> Something like "foobar.clone.123" would be more likely to send me looking
> for "foobar".  (The word "clone" is not very informative there to a
> non-compiler person, something more obviously magical-looking--or just
> "foobar..123" might be better.)

It actually is foo.clone.0 on the testcase in the PR on the common linux
targets and _ZL3fooii.clone.0 in C++.  The first character of _clone
is overwritten with ., $ or kept as _ depending on which characters
are allowed as part of function names on the target.

	Jakub