Re: [cdt-debug-dev] Really Fixing Arrays

["Alain Magloire" <alain@xxxxxxx>]

> 
> On Fri, Jul 18, 2003 at 10:53:23AM -0700, Chris Songer wrote:
> > 
> > Hi,
> > 
> > I'm motivated to fix large arrays and the variable view by doing lazy 
> > retrieval of entries from GDB. My goal is that a user's request for the 
> > last 128 entries of a 32k array create a request to gdb for somewhere 
> > closer to 128 entries than the current request for 32k entries.
> > 
> > The thing I'm not happy about is that -var-list-children does not allow 
> > specification of a range of children. I don't see anything in the MI 
> > documentation or code that lets me get at a variable's children in anything 
> > other than an all or nothing fashion.
> > 
> > Lets presume an array int a[128]. The only thing I can find that will do 
> > what I want is this:
> > 
> > -var-create arraymagicunique_a_1 "*" a[1]
> > -var-evaluate-expression arraymagicunique_a_1
> > 
> > Any other suggestions for how to use MI to do this?
> 
> I'm not up on the MI syntax at the moment, but in the CLI you could do
> something like this: "print a[32734]@32" would give you entries 32734 -
> 32765.  Probably you can use evaluate-expression to do this also.
> 
> An extension to -var-list-children may be in order.
> 

Yes, we've created a PR(GDB gnats) on this, do not have the # handy .. wait ..
GDB # 793


We actually use this for typecasting, for example when casting to array.

But you are right, we can fake the children array range by using
this feature.  I will look into it.

Thanks for the heads up.