Thoughts on saving breakpoints

[Tom Tromey <tromey at redhat dot com>]

Last week I did some debugging and I had an experience which helped me
understand why I think breakpoints should be saved the way I think
they should be saved.

I was debugging gcc, and (thought I :-) found my bug.  I fixed it and
rebuilt gcc without stopping insight.  When I re-ran the inferior, gdb
re-read the gcc executable, recomputed (I presume) and reinserted my
breakpoints, and then actually restarted the debuggee.

My feeling is that saving the Insight session and restarting it should
have roughly the same user-visible effect as recompiling the
inferior's executable.

We all know there's no perfect way to save and restore breakpoints due
to possible changes in the inferior.  However, I think that if we
adopt the same strategy that gdb itself uses when re-reading the
executable, then this will be the least suprising to the user.  I for
one would find it confusing if gdb and Insight adopted different
heuristics here.

Tom