Re: (gdb-6.8) Discard breakpoint address if shared library is unloaded

[Nick Roberts <nickrob@snap.net.nz>]

 > > Since it looks like Vladimir can't find the time to look at this
 > 
 > Huh? I don't think expecting a response within one business day
 > is quite reasonable.

I mean since last November when I first raised the issue.

 >...
 > >  > The idea is that you can't have header_of_multiple=1 and b->loc == NULL
 > >  > at the same time. So it's OK to move the check for header_of_multiple
 > >  > up.  And that avoids the code duplication.
 > > 
 > > I agree.
 > > 
 > > I can confirm that Gdb works as I would expect with these changes.
 >
 > Since you seem to have a patch relative to previous one, would you mind
 > sending a complete patch relative to CVS HEAD?

Joel's version is relative to CVS HEAD.  All I'm suggesting is that also
requires that the line

   ui_out_table_header (uiout, 4, ui_left, "enabled", "Enb");	/* 4 */

is changed back to

   ui_out_table_header (uiout, 3, ui_left, "enabled", "Enb");	/* 4 */

 > > 
 > > On a related note, breakpoint.c now has 7 columns:
 > > 
 > >   ui_out_table_header (uiout, 7, ui_left, "number", "Num");		/* 1 */
 > > 
 > > up from the previous 3:
 > > 
 > >   ui_out_table_header (uiout, 3, ui_left, "number", "Num");		/* 1 */
 > > 
 > > presumably for breakpoint numbers like 1.2 etc but it still seems to
 > > anticipate a very large number of breakpoints.
 > 
 > That's part of the original multiple location patch; I don't think
 > it's related in any way to what we're discussing now.

Granted, it's not central to the issue at hand but it's related in that it
also involves formatting of the header of the breakpoints table and was
made on the same day.

-- 
Nick                                           http://www.inet.net.nz/~nickrob