Add TCP configuration parameters for Transport API.

Add TCP configuration parameters for Transport API.

BUG=91439
TEST=None

Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=99776

[brettw, 2011-09-01 16:49:19]

Since you weren't setting no_delay in this patch, I wonder if there's some way
we can have a test for this class...

http://codereview.chromium.org/7820008/diff/2001/webkit/glue/p2p_transport.h
File webkit/glue/p2p_transport.h (right):

http://codereview.chromium.org/7820008/diff/2001/webkit/glue/p2p_transport.h#...
webkit/glue/p2p_transport.h:66: // Disables Neagle's algorithm when set to true.
Can you expand on this comment to describe what it means. I looked up Neagle's
algorithm and I'm not entirely sure why "no delay" is an appropriate description
of it.

http://codereview.chromium.org/7820008/diff/2001/webkit/plugins/ppapi/ppb_tra...
File webkit/plugins/ppapi/ppb_transport_impl.cc (right):

http://codereview.chromium.org/7820008/diff/2001/webkit/plugins/ppapi/ppb_tra...
webkit/plugins/ppapi/ppb_transport_impl.cc:179: if (value.type !=
PP_VARTYPE_BOOL) {
No {} for single-line (for consistency).

http://codereview.chromium.org/7820008/diff/2001/webkit/plugins/ppapi/ppb_tra...
webkit/plugins/ppapi/ppb_transport_impl.cc:182: config_.tcp_receive_window =
value.value.as_bool;
Copy paste error: you meant to set no_delay.

[Sergey Ulanov, 2011-09-03 21:10:02]

I updated the tests for the API. I'm not sure if there is a good way to verify
that the options are really set without mocking bottom layers of the
implementation, which would be hard to do given that the test runs as a PPAPI
plugin.

http://codereview.chromium.org/7820008/diff/2001/webkit/glue/p2p_transport.h
File webkit/glue/p2p_transport.h (right):

http://codereview.chromium.org/7820008/diff/2001/webkit/glue/p2p_transport.h#...
webkit/glue/p2p_transport.h:66: // Disables Neagle's algorithm when set to true.
On 2011/09/01 16:49:19, brettw wrote:
> Can you expand on this comment to describe what it means. I looked up Neagle's
> algorithm and I'm not entirely sure why "no delay" is an appropriate
description
> of it.

I extended comments for this option in ppb_transport_dev.h. It is conventionally
called 'no delay' (TCP_NODELAY to be exact) in other TCP implementations, e.g.
see "man 7 tcp" (http://en.wikipedia.org/wiki/Nagle's_algorithm also mentions
it)

http://codereview.chromium.org/7820008/diff/2001/webkit/plugins/ppapi/ppb_tra...
File webkit/plugins/ppapi/ppb_transport_impl.cc (right):

http://codereview.chromium.org/7820008/diff/2001/webkit/plugins/ppapi/ppb_tra...
webkit/plugins/ppapi/ppb_transport_impl.cc:179: if (value.type !=
PP_VARTYPE_BOOL) {
On 2011/09/01 16:49:19, brettw wrote:
> No {} for single-line (for consistency).

Done.

http://codereview.chromium.org/7820008/diff/2001/webkit/plugins/ppapi/ppb_tra...
webkit/plugins/ppapi/ppb_transport_impl.cc:182: config_.tcp_receive_window =
value.value.as_bool;
On 2011/09/01 16:49:19, brettw wrote:
> Copy paste error: you meant to set no_delay.

Done.

[brettw, 2011-09-06 18:17:00]

LGTM