Re: [RFA take 4] Allow setting breakpoints on inline functions (PR 10738)

[Gary Benson <gbenson@redhat.com>]

Doug Evans wrote:
> On Wed, Feb 15, 2012 at 1:47 AM, Gary Benson <gbenson@redhat.com> wrote:
> > GDB's behaviour is inconsistent if you don't rejecting the old
> > index files.  That seemed like a bad thing to be introducing.
> 
> Absent a notification to the user and ability to control it, sure.
> But I think we can come up with something appropriate.

Ok.

> > > One could support the old version for a release or two, and
> > > print a warning when older versions are encountered.
> >

> > I wondered about this myself, though it was pointed out to me that
> > printed warnings often get lost in the noise.
> 
> One can debate this forever. :-)

Well, quite :)

> > > The user's build procedure may involve building the index in a
> > > way that is not easily updated in a timely manner.  Thus all the
> > > speed improvements are (at least temporarily, but for a long
> > > enough time to be troublesome) wiped out simply by using a
> > > *newer* version of gdb.  And that makes me uncomfortable.
> >
> > How would it be if there the default behaviour was be to reject
> > old indexes (as the patch does now) but with the addition of a
> > flag ("maint set allow-old-gdb-indexes" perhaps?) that would allow
> > users in this particular situation to get around it?  That way,
> > our response to complaints can be "rebuild the index *or* use this
> > flag (which by the way will lose you such and such a
> > functionality)"  Inconsistent behaviour doesn't seem so bad if the
> > user asked for it.
> 
> IWBN if one could do "gdb my-binary-with-older-index" (as opposed
> to, e.g., "gdb ; maint set ... ; file my-binary-with-older-index",
> or the equivalent with -x/-ex foo, and setting the flag in
> ./.gdbinit won't work).  That pretty much means passing gdb an
> option ("gdb --use-old-index my-binary" or some such).  At Google
> we've added --disable-gdb-index as an escape hatch against broken
> indices.  I'm happy to replace it with something that will do the
> same thing.
> 
> As for what the default behaviour should be, I don't have a strong
> enough opinion to want to defend it.  I can easily enough change it
> here if desired.  [Not something unfamiliar to Redhat. :-)]

Ok, so my plan is to implement an --allow-incomplete-gdb-index option
that will turn on support for version 4 and 5 indexes.  I am unsure
as to whether to add printed warnings, but I'm leaning towards not
having warnings in either configuration.  With the default setting
GDB will work correctly, but may possibly be slow which would be the
(admittedly not very direct) trigger for the user to investigate.
With the option supplied I assume the user knows what they're doing,
so they don't need pestering!  I like the *idea* of warning users
that their .gdb-index sections are being skipped, but I think of the
case where somebody uses a new GDB on an older version of Fedora (say)
being flooded with warning messages about every single shared library
their program is linked to.

Cheers,
Gary

-- 
http://gbenson.net/