Re: [PATCH] mi-symbol-*

[Jelmer Vernooij <jelmer@nl.linux.org>]

On Wed, Nov 06, 2002 at 08:56:23AM +0200, Eli Zaretskii wrote about 'Re: [PATCH] mi-symbol-*':
> The documentation patch is approved, but please take care of these minor 
> problems before it is committed:
>
>  - Please break all lines inside @smallexample that are longer than 64 
>    characters, otherwise they will produce warnings from TeX and ugly 
>    results in the printed manual.  (You can tell in a note that the 
>    broken line is one long line.)
> 
>  - Why did you add so many empty lines?  IMHO, this produces unpleasant 
>    results, both in the on-line (Info and HTML) and printed formats.
> 
>  - Where you say "pattern", please add that the pattern is a regular 
>    expression.
All fixed; is the update ok?

On Thu, Nov 07, 2002 at 11:59:52AM -0500, Elena Zannoni wrote about 'Re: [PATCH] mi-symbol-*':
> Info_line should return the core addr ranges for a given line number.
> You instead have implemented a different behavior: given a line number
> name, print the line number which we just gave as input, and the file name.
> Also in the usage and error messages, the use of the word 'symbol' is a bit
> inappropriate, because the argument of the command is not a symbol name.
Is there any documentation about the internals of gdb and what the
various terms mean? I do have some knowledge of how executables are
structured, but I don't know any details or specific terms.

>  > +enum mi_cmd_result
>  > +mi_cmd_symbol_info_function (char *command, char **argv, int argc)
>  > +{
>  > +  char *function = "";
>  > +  struct symtabs_and_lines sals;
>  > +  struct symbol *sym;
>  > +  struct minimal_symbol *msym;
>  > +  struct cleanup *old_chain;
>  > +  char *args, **canonical;
>  > +
>  > +  if (argc != 1)
>  > +    error ("mi_cmd_symbol_info_function: Usage: SYMBOL");
>  > +
>  > +  args = argv[0];
>  > +
>  > +  sals = decode_line_1 (&args, 1, NULL, 0, &canonical);
> 
> this one could be ok, but there is contradictory information in the
> current documentation. It was documented as taking a symbol and
> returning the function where the symbol is in, however it suggest that
> it should provide the functionality that a gdbtk function
> provides. That gdbtk function however takes a linespec as argument.
> 
> I think this mi command needs to be revisited.
> So, wait until the issue is cleared.
Any news on the interface?

Thanks,

Jelmer