Change ParseHostAndPort() to not include brackets around IPv6 literals.

Change ParseHostAndPort() to not include brackets around IPv6 literals.

There were several consumers assuming that the returned hosts had no brackets.

BUG=417417
Committed: https://crrev.com/ff374b7fd49f6ddf9f4ee597759d65996534379a
Cr-Commit-Position: refs/heads/master@{#296569}

[eroman, 2014-09-24 21:43:34]

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

[Ryan Hamilton, 2014-09-24 21:48:49]

lgtm. did you check that --host-resolver-rules now does the right thing?

https://codereview.chromium.org/602973002/diff/1/net/proxy/proxy_bypass_rules.cc
File net/proxy/proxy_bypass_rules.cc (right):

https://codereview.chromium.org/602973002/diff/1/net/proxy/proxy_bypass_rules...
net/proxy/proxy_bypass_rules.cc:310: std::string bracketed_host =
HostPortPair(host, 80).HostForURL();
nit: s/80/port/ ?  (I suspect it doesn't actually matter, but still :>)

[Ryan Hamilton, 2014-09-24 21:49:05]

Oh, and THANKS for the super quick fix!

[eroman, 2014-09-24 21:59:28]

Yes I did test that the command line flag is fixed (you need to us the bracketed
version).

https://codereview.chromium.org/602973002/diff/1/net/proxy/proxy_bypass_rules.cc
File net/proxy/proxy_bypass_rules.cc (right):

https://codereview.chromium.org/602973002/diff/1/net/proxy/proxy_bypass_rules...
net/proxy/proxy_bypass_rules.cc:310: std::string bracketed_host =
HostPortPair(host, 80).HostForURL();
On 2014/09/24 21:48:49, Ryan Hamilton wrote:
> nit: s/80/port/ ?  (I suspect it doesn't actually matter, but still :>)

The specific port used does not matter, I am just using that function in order
to potentially bracket IPv6 literals. (Will document).

I considered changing ParseHostAndPort() to returning a HostPortPair, which
would make IPv6 behavior more clear (and in fact why HostPortPair was introduced
in the first place).

The issue with that is that HostPortPair makes ports required, whereas
ParseHostAndPort considers the port optional (by setting to -1 when
unspecified).

... Long story short, I hardcoded port=80 because I don't want to pass -1 into
HostPortPair's constructor, which expects a uint16. Because integer underflow is
not cool.

[eroman, 2014-09-24 22:55:32]

The CQ bit was checked by eroman@chromium.org

[commit-bot: I haz the power, 2014-09-24 22:56:28]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/602973002/20001

[commit-bot: I haz the power, 2014-09-24 23:42:07]

Committed patchset #2 (id:20001) as a8528a560d257469365d3a4eb7cf2579e5296e0e

[commit-bot: I haz the power, 2014-09-24 23:47:15]

Patchset 2 (id:??) landed as
https://crrev.com/ff374b7fd49f6ddf9f4ee597759d65996534379a
Cr-Commit-Position: refs/heads/master@{#296569}