Re: Questions about QPACK

[Dmitri Tikhonov <[e-mail]>]

On Sat, Nov 23, 2019 at 01:23:41AM +0800, 漂泊的梦 wrote:
> Thanks for your reply. There is some misunderstanding about Q2.

To quote Q2 from the original email:

>> https://tools.ietf.org/html/draft-ietf-quic-qpack-11#section-4.5.4
>> 
>>         0   1   2   3   4   5   6   7
>>       +---+---+---+---+---+---+---+---+
>>       | 0 | 1 | N | T |Name Index (4+)|
>>       +---+---+---+---+---------------+
>>       | H |     Value Length (7+)     |
>>       +---+---------------------------+
>>       |  Value String (Length bytes)  |
>>       +-------------------------------+
>> 
>> Q2:when we encode new name, new value, we should insert name,value
>> pair into dynamic table, should we insert name only into dynamic
>> table?

The answer to "should" is "depends on what you want to do."  (See below).

> The background is that when the sender inserts the pair of name &
> value into dynamic table, it should also send the pair of name &
> value and index to recipient through encoder stream.  In case
> that the pair of name & value cannot be found, we can search by
> name only.  The question is that in order to search by name only,
> can we insert name only (without value or with 0-length value)
> and index into the dynamic table, and then send the name and index to
> recipient through encoder stream?  If we don't insert name only and
> index into the dynamic table, how can we search by name only?

Yes, it is valid to insert into the dynamic table an entry with 0-length.
This is useful when the header name repeats, but the value does not.  In
that case, you can reference the dynamic table from the header block via
the Literal Header Field With Name Reference instruction (listed in the
original Q2 above).

  - Dmitri.