Re: [Libunwind-devel] documentation for unw_init_local_signal

[Doug Moore]

I don't object to the proposed change.

There does remain a mysterious difference between  
uwn_init_local_signal and unw_init _remote - namely, the setting of  
c->validate=0 in one and not the other.  It did lead to a difference  
in outcome when I tried to use unw_init_remote to address my original  
problem.  I can imagine that someone, someday, will want a version of  
initialization that sets c->validate to 1 for some reason.  So a  
little extensability makes sense.

Doug

Quoting Dave Watson <address@hidden>:

> On 04/11/17 03:05 PM, Doug Moore wrote:
> > If there was to be a new man page for unw_init_local_signal, there  
> > would be a reason to add another ‘see also’ to the page for  
> > unw_init_remote.  But no new page, so no need.
> >
> > The modified unw_init_local page is good enough.  It just wasn’t  
> > what I was expecting.
>
> So I've still been mulling this fix over.  I'm tempted to change the  
> interface to:
>
> unw_init_local_ex(unw_cursor_t, unw_context_t, int flags);
>
> with a UNW_SIGNAL_FRAME flag - to try to future proof the interface a
> bit.  Or possibly even farther,
>
> unw_init(unw_cursor_t, unw_addr_space_t, void* arg, int flags);
>
> I think might work too.
>
> Any objections or suggestions?
>
> !DSPAM:10223,58ee4c4439341776378724!