Re: RFC: introduce struct ksymbol

[Frederic Weisbecker <fweisbec@gmail.com>]

On Fri, Apr 17, 2009 at 12:55:33AM -0700, Joe Perches wrote:
> On Wed, 2009-04-15 at 20:21 +0930, Rusty Russell wrote: 
> > On Wed, 15 Apr 2009 03:28:39 pm Ingo Molnar wrote:
> > > Why not 'struct ksym'? That name is unused right now, it is shorter 
> > > and just as descriptive.
> > > 
> > > Regarding the change... dunno. Sam, Rusty - what do you think?
> > 
> > Yes, ksym is nice.  But agree with you that it's marginal obfuscation
> > to wrap it in a struct.
> > 
> > The current symbol printing APIs are awful; we should address them first
> > (like the %pF patch does) IMHO.
> 
> I suggest just %pS<type>
> 
> With %pS, struct ksym is probably not all that
> useful unless it's for something like a sscanf.
> 
> Today there are these symbol uses:
> name, offset, size, modname
> 
> So perhaps %pS<foo> where foo is any combination of:
> 
> n name
> o offset
> s size
> m modname
> a all
> 
> and if not specified is a name lookup ("%pSn").


Joe,

It seems to me a rather good idea, it offers a good granularity
about what has to displayed.

The only problem is the end result:

%pSnosm, %pSno, %pSosm, ...

One could end up stuck reading such a format, trying
to guess if the developer wanted to print the symbol +
"nosm" or something...

But since I don't see any point in printing nosm directly after
a symbol... :)

I like this.

Anyone? Any doubt?



> 
>