Re: [hybi] method of allocation of reserved bits to extensions

[Andy Green <[e-mail]>]

On 04/28/2011 11:04 AM, Somebody in the thread at some point said:
> On 28 April 2011 12:13, John Tamplin<[e-mail]>  wrote:
>> reserved bits, it seems reasonable to simply define that the Foo extension
>> allocates RSV2.  If another extension also statically allocates RSV2, then
>> obviously they couldn't be used together.
>
> It might be reasonable if you are google, as you will have the market
> share such that any decision you make to allocated a reserved bit for
> your own extensions will effectively allocate that bit to you forever.
>
> Imagine I'm an intermediary vender and I develop a private extension
> for my WS aggregator to talk to the servers behind it that uses a
> reserve bit.    If google chrome subsequently comes out with a very
> popular extension that uses the same reserved bit, then my
> intermediary product will have been badly impacted.  I will either
> have to redeploy all my intermediaries using another reserved bit or
> lose market share because I don't support googles latest extension.
>
> it is not a level playing field.

It makes me squiffy to think how little one can look at this traffic and 
know what it is by eye, but I have to agree your logic for reserved bit 
and reserved opcode allocation for non-control opcodes seems pretty 
sound.  Stuff won't play well together without it if it wants to use 
them, and x-google-mux does want an opcode at least.

You're right John can hope to mandate that and we would be dead in the 
water doing the same with out own extensions, having to come cap in hand 
begging for a bit or opcode that we already is a jealously guarded 
resource.  But if they are dynamically allocated everyone can exploit 
them without central control or politics.

-Andy