Re: HPACK opcode bit patterns

On 2014–08–06, at 8:13 AM, Osama Mazahir <OSAMAM@microsoft.com> wrote:

>  3) 001xxxxx: "Maximum Header Table Size Change"
> 4) 0001xxxx: "Literal Header Field Never Indexed"
> 5) 0000xxxx: "Literal Header Field Without Indexing”

>  
> Do we really expect tablesize (3) to be more common than literal_without_indexing (5)?

The XXXX in 4-5 is only used to refer to something that’s already been indexed, without bumping it to the front of the index table. Therefore it is unusual for XXXX to be nonzero, regardless of how often cases 4-5 happen overall. (Also, note that the overhead of overflow is just one byte.)

> I don’t have a strong opinion to change the masks, but I am curious if the bit pattern selection was deliberate.


Also, an implementation can treat “without indexing” as “never indexed” and use a common code path, since the encodings are identical modulo a don’t-care bit. Wouldn’t be surprised if that was deliberate.