Re: [PATCH] MI: Free values when updating

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

On Wednesday 24 January 2007 12:14, Nick Roberts wrote:
>  > >  > @@ -1987,11 +1987,7 @@ c_value_of_root (struct varobj **var_han
>  > >  >        /* We need to catch errors here, because if evaluate
>  > >  >           expression fails we just want to make val->error = 1 and
>  > >  >           go on */
>  > > 
>  > > This comment is not applicable anymore.
>  > 
>  > It actually is -- the comment says why we use gdb_evaluate_expression,
>  > as opposed to evaluate_expression. Only the part about val->error is obsolete,
>  > and I'll fix that.
> 
> I'm not even sure that current use of gdb_evaluate_expression with variable
> objects is sensible.  Currently GDB accepts:
> 
>   -var-create - * 1/0
> 
> and if you do:
> 
>   -var-create - * n1/n2
> 
> and n2 is set to 0,  with "-var-update --all-values", GDB returns:
> 
>   ^done,changelist=[{name="var2",in_scope="false"}]

Yes. in_scope="false" actually means 
"there's being some kind of error evaluating the expression". It's not very
clear but not a big problem either. It might be good to clean this up, 
but I'm nore interested in fixing the (IMO) large scoping problems --
as I've suggested in -var-list --all-locals proposal.

- Volodya