Re: RFC patch mi-cmds.c table format (pedantic)

[Jason Molenda <jmolenda@apple.com>]

On Mar 18, 2004, at 11:55 AM, Andrew Cagney wrote:

> First, have a look at:
> http://sources.redhat.com/cgi-bin/gnatsweb.pl?cmd=view%20audit- 
> trail&database=gdb&pr=496

Ah, I hadn't seen that.

How do you envision the cli-mapped mi commands working (e.g. -gdb-set),  
where files in the gdb/ directory are registering the commands?


>> the mi-cmds table changed from this format:
>>   {"break-info", "info break %s", 0},
>>   {"break-insert", 0, 0, mi_cmd_break_insert},
>> to this format:
>>   { "break-info", { "info break", 1 }, NULL, NULL },
>>   { "break-insert", { NULL, 0 }, 0, mi_cmd_break_insert},
>> As you can see here, the use of 0 vs NULL is inconsistent,
>
> That's cos it was converted using an emacs macro.


Emacs?  Ah, well, I think we can all see the problem right there. :-)

>> ok, but perhaps don't bother with the trailing initializers.

OK I'll come up with a real/tested patch.  The trailing initializers --  
you don't mean this kind of change, do you?

>> -  { "exec-step", { NULL, 0 }, mi_cmd_exec_step},
>> +  { "exec-step", { NULL, 0 }, mi_cmd_exec_step, NULL},

I think you were talking about the final NULL entry.


> Should I ask what motived this?

Our GUI guys asked for -exec-abort to be implemented so I had cause to  
look at our tasty-fresh-mid-merge sources for the first time, where  
we're just picking up FSF gdb changes from the last year or so.


J