Re: A patch I'd like to propose.

[Pedro Alves <pedro at codesourcery dot com>]

On Wednesday 14 April 2010 15:29:24, Chris Moller wrote:
> On 04/14/10 10:19, Phil Muldoon wrote:
> > On 04/14/2010 12:59 PM, Chris Moller wrote:
> >    
> >> Adds the command "quickquit," aliased to "qq," to cli/cli-cmds.c.   It
> >> eliminates that annoying quit_confirm() when quitting gdb.
> >>      
> >
> > 'set confirm off' works for me?  Though this is not the default it
> > seems.  Is this orthogonal to your request, or did you want an
> > explicit command to act regardless of confirm off/on?
> >    
> 
> I'm looking for something that always works without me having to make 
> any special arrangement for it.  I use a lot of bizarre environments in 
> debugging and a lot of the time I've forgotten things like copying over 
> my usual~/.gdbinit

That's what you should be fixing then, isn't it?  ;-)

 define qq
   set confirm off
   quit
 end

Kinda misses the scripting point we add commands for everything
that's _already_ easily possible, even with CLI scripting.

In any case, I'd be okay with this with the following constrains:

* It was rather called "quit-force", or something similarly, so that:

 - q<tab> does't stop completing to "quit", but instead even
   proposes "quit"/"quit-force" to the user.  As bonus, users notice
   the new command is available.

 - the fact that `qq' is typo-prone is avoided.  E.g., "quit-force" -> qf.

* q, qu, qui, remain unambiguous aliases to "quit".  "quickquit" shares
3 initials with "quit", so q will stop working if aliases are not added.

* You explain whether you mean for this to disable _all_ queries
when quitting or not.  I think that's what you meant.  As is, it
doesn't do that, because e.g.,

 +void
 +quick_quit_command (char *args, int from_tty)
 +{
 +  disconnect_tracing (from_tty);
    ^^^^^^^^^^^^^^^^^^

This can query.

 +
 +  quit_force (args, from_tty);
 +}

* Of course, that this is documented in the manual.

-- 
Pedro Alves