Hi Florian,
On Wed, Nov 06, 2024 at 03:32:53PM +0100, Florian Westphal wrote:
> Florian Westphal <fw@xxxxxxxxx> wrote:
> > > From userdata path it should be possible to check for this special
> > > internal queue_datatype then encode the queue number type in the TLV.
> >
> > I have no idea how to do any of this. I don't even know what a "queue number
> > type" is.
> >
> > How on earth do i flip the data type on postprocessing without any idea
> > what "2 octets worth of data" is?
>
> You seem to dislike EXPR_TYPE; I tried to sketch something but i would
> turn EXPR_VALUE into EXPR_TYPE, including EXPR_TYPEOF_NFQUEUE_ID and
> the udata build/parse functions, with the addition of a
>
> /* Dummy alias of integer_type for nf_queue id numbers */
> const struct datatype integer_queue_type = {
>
> that has no actual function except to override what constant_expr_print()
> ends up doing.
I am fine with your patch, it is perfectly fine to address this in
this way. I just proposed a different way to handle this special case.
I can take a look later today based on your patch, I think I can reuse
90% of it, it is just a subtle detail what I am referring to.
