Re: [RFA/RFC] Tweak for a gdb.mi test.

[Michael Snyder <msnyder@redhat.com>]

Andrew Cagney wrote:
> 
> > On Tue, May 07, 2002 at 06:09:11PM -0700, Michael Snyder wrote:
> >
> >>
> >> I'm gonna ask for a second pair of eyes, since I don't know MI
> >> very well.
> >>
> >> What this is -- the test is examining the stack, but it is
> >> assuming that main is the last frame.  My change allows for
> >> one extra frame below main (eg. for '_start').
> >>
> >> OK to check in?
> >
> >
> > Before you check this in, I would prefer to have a policy decision
> > in place about whether we should show that frame or not.  The relevant
> > macro is FRAME_CHAIN_VALID; I believe we should universally (or almost
> > universally) change this to stop at main.  I think that's
> > func_frame_chain_valid but don't trust my memory.
> 
> (don't remember which function either, but)
> Yes, I don't think the backtrace should go past main so I think the
> change is wrong.
> 
> I remember much debate about the test at the time (it was Cygnus
> internal unfortunatly).  The thing that clinched the deal was the
> obeservation (made by a human factors person) that the behavour had to
> be consistent across platforms. 

1) It may have to be, but it isn't.
2) Why does it have to be?  Lots of behaviors differ between platforms.
3) My real reason for submitting this fix is that it breaks with the
"needs_status_wrapper" change that I submitted the other day.
Of course, technically it would not break if gdb always stopped the
trace at main, but it doesn't...


> For a given OS (e.g. eCos, GNU/Linux,
> ...) the backtrace should look identical, regardless of the ISA.
> 
> Having each ISA making independant, and somewhat arbitrary, decisions is
> wrong.

Whether that's so or not, they do.  Do you want to have this
always fail for current gdb ports?


>  From memory, a suggestion was to let people select the back-trace
> policy independant of the current architecture.

I thought we also had a policy of not inserting tests
that we knew would fail on some targets?  Something about
this being a regression test...

Sorry, don't mean to be snippy...