Re: Re-work of op-code patterns

On Mon, Jan 27, 2014 at 9:38 AM, Martin Thomson <martin.thomson@gmail.com>wrote:

> On 27 January 2014 08:43, Jeff Pinner <jpinner@twitter.com> wrote:
> > With the proposed change what would 0b1000000 signal?
>
> An indexed header field, the first one.
>
> Indexes start at 1, so 0x80 is not a valid indexed representation -- that
is why it was chosen for "clear reference set." To be clear, my proposal is
that state changes to the table (a rare condition) and any padding (a
purposely increased compressed size) are signaled by 0x80 followed by a new
set of opcodes. This requires no changes to the existing opcodes and incurs
no penalty to compression when not modifying the state.