Routing to net for triage. It is unlikely that these are Mac-specific.

> It is unlikely that these are Mac-specific.

Yes, it's not platform-specific. Just peeked Mac, since as far as I remember platform was a required field in wizard.

I'll take a look at this.

A lot of these are us deliberately copying Firefox's quirks.  In cases where other browsers get away with more closely adhering to the spec, we can presumably fix behavior, but in cases where all browsers behave in the same way, it may be better to make the spec reflect what all browsers do, just from a not breaking the world standpoint, unfortunate as it may be.

Agreed overall, but these reported cases include only one where Chrome and Firefox behave identically to each other and differently from the majority of tested browsers. In the remaining four reported cases Chrome behaves differently from all the other tested browsers.

As bsittler says, this report is only for the few cases that are unique to Chrome and Firefox. The full chart is fascinating: https://inikulin.github.io/cookie-compat/

True.  The ones that I think we may have to keep are the "Set-Cookie: blah" ones, though Safari's behavior is different enough there (At least for the "Set-Cookie: blah" / "Set-Cookie: JumboShrimp" case, that maybe we don't)

I've uploaded https://codereview.chromium.org/2246613003/ to hopefully address the listed issues.

A quick clarification, however. Perhaps I'm missing it on the chart, but should all attributes be treated the same as domain in OPTIONAL_DOMAIN0042? That is, for example, if I have:
  Set cookie: 
  "foo=bar; expires=Sun, 18-Apr-2027 21:06:29 GMT; expires="
should we ignore the second expires?

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/03e6ff8c614040f60d1a1d818411870ddb3b90e4

commit 03e6ff8c614040f60d1a1d818411870ddb3b90e4
Author: jww <jww@chromium.org>
Date: Wed Aug 17 19:19:58 2016

Update cookie value and attribute setting behavior to match other UAs

This is an attempt to bring Chrome's cookie behavior in line with the
majority of other major UAs, and thus are de facto standards. The
specific updates are:
  1. An empty domain attribute value should be ignored as a
     non-attribute rather than setting an empty attribute value.
  2. An empty cookie line (or a cookie line with ignored characters)
     should be treated as setting an empty value for the empty-key.

BUG= 601786 

Review-Url: https://codereview.chromium.org/2246613003
Cr-Commit-Position: refs/heads/master@{#412608}

[modify] https://crrev.com/03e6ff8c614040f60d1a1d818411870ddb3b90e4/chrome/browser/browsing_data/browsing_data_cookie_helper_unittest.cc
[modify] https://crrev.com/03e6ff8c614040f60d1a1d818411870ddb3b90e4/net/cookies/cookie_store_unittest.h
[modify] https://crrev.com/03e6ff8c614040f60d1a1d818411870ddb3b90e4/net/cookies/parsed_cookie.cc
[modify] https://crrev.com/03e6ff8c614040f60d1a1d818411870ddb3b90e4/net/cookies/parsed_cookie.h
[modify] https://crrev.com/03e6ff8c614040f60d1a1d818411870ddb3b90e4/net/cookies/parsed_cookie_unittest.cc