Re: [PATCH nft v3 2/2] datatype: Implement binary search in symbolic_constant_print()

[Florian Westphal <fw@strlen.de>]

Pablo Neira Ayuso <pablo@netfilter.org> wrote:
> On Mon, Nov 28, 2016 at 11:12:23PM +0100, Florian Westphal wrote:
> > This is unmaintanable.  I have no clue what value ETH_P_8021Q is, and that
> > this has to be placed at spot #2 to not break things.
> 
> OK, let's try to make this a bit more maintainable, another proposal
> in steps:
> 
> 1) Update symbol_table definition to:
> 
> struct symbol_table {
>         unsignd int                     size;
>         const struct symbolic_constant  *symbols; <---
> };

[..]

> Florian, are you OK if Elise follows this approach? If this is overly
> complicated and not worth, rise your hand.

Is a linear search of ~300-element symtable list a problem?

If it is, then your proposal seems like a sane approach.
But I'd probably benchmark this first.

(Since Elise already did a patch for bsearchable symtable
 it should not be too much work to verify this).