This is the mail archive of the
gdb-patches@sourceware.org
mailing list for the GDB project.
Re: [RFA] Reference value coercion
- From: Paul Hilfinger <hilfingr at gnat dot com>
- To: gdb-patches at sourceware dot org
- Date: Sat, 20 May 2006 06:10:14 -0400 (EDT)
- Subject: Re: [RFA] Reference value coercion
- Reply-to: Hilfinger at adacore dot com
Daniel,
> The patch is not right for C++. It works for your test, because Child
> has a single base class, and therefore the values of the Child& and
> Parent& are the same; but in more complicated cases, this is no longer
> true. The reference needs to be adjusted.
Yeah. Somehow I knew it couldn't be that simple.
> Also, there's some trouble with the test case: using "runto" restarts
> the inferior, so runto_main followed by runto is redundant. I think
> you just want to use runto. The step_for_stub thing is ancient, and
> I don't think the testsuite works on systems that would require it any
> more.
Umm. Well then you might want to look at ref-types.exp, from which I
stole most of this code.
...
> Does the attached work for Ada?
Yes, thanks, it does.
PNH