Re: [PATCH] Avoid breakpoint query in MI

[Vladimir Prus <ghost@cs.msu.su>]

Nick Roberts wrote:

> This updates my earlier patch
> (http://sourceware.org/ml/gdb-patches/2008-01/msg00421.html) from the thread
> "bug in mi when setting breakpoint" started last year by Dodji Seketeli.  It is
> simplified by Joel's recent patch for multiple symbols and Vladimir's concept
> of top-level interpreter.
> 
> I see the default for multiple-symbols is 'all' (it used to be 'ask', perhaps
> this change should be mentioned in NEWS) which might make this change more
> acceptable.
> 
> Nick                                           http://www.inet.net.nz/~nickrob
> 
> 
> 2008-04-18  Nick Roberts  <nickrob@snap.net.nz>
> 
>         * interps.c (interp_top_level): New function.
> 
>         * interps.h: New extern.
> 
>         * linespec.c: Include interps.h and mi/mi-cmds.h.
>         (decode_line_2): When using MI, always set all breakpoints in menu.
> 
> 
> *** interps.h   15 Mar 2008 18:14:50 +1300      1.14
> --- interps.h   18 Apr 2008 00:47:56 +1200      
> *************** extern int current_interp_display_prompt
> *** 66,71 ****
> --- 66,72 ----
>   extern void current_interp_command_loop (void);
>   /* Returns opaque data associated with the top-level interpreter.  */
>   extern void *top_level_interpreter_data (void);
> + extern void *interp_top_level (void);

Please name this top_level_interpreter. Also, the return type
should be "struct interp*". In fact, I have a not-yet-submitted
patches which does just that :-)


>   
>   extern void clear_interpreter_hooks (void);
>   
> 
> *** interps.c   15 Mar 2008 18:14:50 +1300      1.23
> --- interps.c   16 Apr 2008 15:37:01 +1200      
> *************** interpreter_completer (char *text, char 
> *** 476,482 ****
>     return matches;
>   }
>   
> ! extern void *
>   top_level_interpreter_data (void)
>   {
>     gdb_assert (top_level_interpreter);
> --- 476,488 ----
>     return matches;
>   }
>   
> ! void *
> ! interp_top_level (void)
> ! {
> !   return top_level_interpreter;  
> ! }
> ! 
> ! void *
>   top_level_interpreter_data (void)
>   {
>     gdb_assert (top_level_interpreter);
> 
> 
> *** linespec.c  18 Apr 2008 00:10:58 +1200      1.75
> --- linespec.c  18 Apr 2008 00:21:29 +1200      
> ***************
> *** 36,41 ****
> --- 36,43 ----
>   #include "linespec.h"
>   #include "exceptions.h"
>   #include "language.h"
> + #include "interps.h"
> + #include "mi/mi-cmds.h"
>   
>   /* We share this one with symtab.c, but it is not exported widely. */
>   
> *************** See set/show multiple-symbol."));
> *** 524,530 ****
>   
>     /* If select_mode is "all", then do not print the multiple-choice
>        menu and act as if the user had chosen choice "1" (all).  */
> !   if (select_mode == multiple_symbols_all)
>       args = "1";
>     else
>       {
> --- 526,533 ----
>   
>     /* If select_mode is "all", then do not print the multiple-choice
>        menu and act as if the user had chosen choice "1" (all).  */
> !   if (select_mode == multiple_symbols_all
> !       || ui_out_is_mi_like_p (interp_ui_out (interp_top_level ())))
>       args = "1";
>     else
>       {

I think that adding breakpoints on all symbols is better than silently grabbing 
random symbol, so the above seems fine to me. Can somebody approve the linespec.c
bit?

- Volodya