relnote: Add QUIC connection option DHDT to disable HPACK dynamic table. Guarded by FLAGS

relnote: Add QUIC connection option DHDT to disable HPACK dynamic table. Guarded by FLAGS_quic_disable_hpack_dynamic_table.

Will be used in a Finch trial to gather data about HPACK compression
efficiency: how much of HPACK's efficiency derives from the dynamic
table,
as opposed to other parts of HPACK such as the static table and
huffman coding?

If it is very low then next step would be to:
     o with dynamic table disabled, disable the monolithic header
       stream, to gather data about whether reducing HOL improves other metrics.
     o if results are positive, it has the advantage of maintaining
       (restricted) conformance to the HPACK spec.

Otherwise, some (non-complient) extensions or modifications to HPACK
might be necessary to reduce eliminate headers HOL blocking.

     o one idea involve maintining a small number of separate HPACK
       contexts, to ensure that dynamic reference only occur to older
       entries that are free of HOL blocking potential.

 Merge internal change: 118999202

BUG=

[Zhongyi Shi, 2016-04-08 05:13:53]

Description was changed from

==========
relnote: Add QUIC connection option DHDT to disable HPACK dynamic table. Guarded
by FLAGS_gfe2_reloadable_flag_quic_disable_hpack_dynamic_table.

Will be used in a Finch trial to gather data about HPACK compression
efficiency: how much of HPACK's efficiency derives from the dynamic
table,
as opposed to other parts of HPACK such as the static table and
huffman coding?

If it is very low then next step would be to:
     o with dynamic table disabled, disable the monolithic header
       stream, to gather data about whether reducing HOL improves other metrics.
     o if results are positive, it has the advantage of maintaining
       (restricted) conformance to the HPACK spec.

Otherwise, some (non-complient) extensions or modifications to HPACK
might be necessary to reduce eliminate headers HOL blocking.

     o one idea involve maintining a small number of separate HPACK
       contexts, to ensure that dynamic reference only occur to older
       entries that are free of HOL blocking potential.

 Merge internal change: 118999202

BUG=
==========

to

==========
relnote: Add QUIC connection option DHDT to disable HPACK dynamic table. Guarded
by FLAGS_quic_disable_hpack_dynamic_table.

Will be used in a Finch trial to gather data about HPACK compression
efficiency: how much of HPACK's efficiency derives from the dynamic
table,
as opposed to other parts of HPACK such as the static table and
huffman coding?

If it is very low then next step would be to:
     o with dynamic table disabled, disable the monolithic header
       stream, to gather data about whether reducing HOL improves other metrics.
     o if results are positive, it has the advantage of maintaining
       (restricted) conformance to the HPACK spec.

Otherwise, some (non-complient) extensions or modifications to HPACK
might be necessary to reduce eliminate headers HOL blocking.

     o one idea involve maintining a small number of separate HPACK
       contexts, to ensure that dynamic reference only occur to older
       entries that are free of HOL blocking potential.

 Merge internal change: 118999202

BUG=
==========

[Zhongyi Shi, 2016-04-08 05:13:53]

zhongyi@chromium.org changed reviewers:
+ rch@chromium.org

[Zhongyi Shi, 2016-04-11 17:36:37]

zhongyi@chromium.org changed reviewers:
+ rtenneti@chromium.org

[ramant (doing other things), 2016-04-11 18:17:49]

lgtm

https://codereview.chromium.org/1865803007/diff/1/net/quic/quic_flags.cc
File net/quic/quic_flags.cc (right):

https://codereview.chromium.org/1865803007/diff/1/net/quic/quic_flags.cc#newc...
net/quic/quic_flags.cc:43: // HPACK\'s dynamic table.
nit: Why back slash?

https://codereview.chromium.org/1865803007/diff/1/net/tools/quic/end_to_end_t...
File net/tools/quic/end_to_end_test.cc (right):

https://codereview.chromium.org/1865803007/diff/1/net/tools/quic/end_to_end_t...
net/tools/quic/end_to_end_test.cc:196: // Run version negotiation tests tests
with no options, or all
overly nit: should "tests tests" be "tests"?