Just to note the complexity, lukasza@ pointed out that this risks resetting SitePerProcess for users who had it enabled before.  We'll want to see if we can map current Enabled values to Enabled in the tri-state, and current Disabled values to Default in the tri-state.  Not sure if that's easy or not.

This is a bit tricky, unfortunately, since we don't want Chrome to forget if a user enabled the value before.  I've confirmed that switching from SINGLE_VALUE_TYPE to FEATURE_VALUE_TYPE doesn't preserve site-per-process if the user had it before.  Same for switching from SINGLE_VALUE_TYPE to ENABLE_DISABLE_VALUE_TYPE, since the enabled value changes from "enable-site-per-process" to "enable-site-per-process@1" and doesn't match.

Alexei suggested looking at custom code when the value is read from disk to enable it.  Nick is considering how to add support for this with a migration rule, such that we can switch over to a FEATURE_VALUE_TYPE and set it to enabled if the SINGLE_VALUE_TYPE had been set before.

Nick said he can probably take a look after his current CL.  Let's try to target this for next week's Dev Channel build if possible.  Thanks!

During "upgrade", you could just directly read the pref "browser.enabled_labs_experiments" and see if this has been set or not. You can see this in chrome://local-state

Fix landed in r542288.  (Description below, since it was missing the bug number and didn't trigger an update.)  I've verified it works in Chrome Mac Canary 67.0.3368.0.

govind@, ok if we merge this to M66 for beta?  We'd like to use it to help users with field trials.

---

https://chromium-review.googlesource.com/c/chromium/src/+/947398

Add a flag that has the effect of force-disabling the site isolation field
trial.

It's a flag, rather than a Feature, so that it can be marked as unsafe later on.

(Ideally we would make the existing #enable-site-per-process flag multi-state,
but there appears to be no ready way to do that without losing the old setting:
there's no migration scheme at present, and since flags storage isn't
versioned, all the ways I can see to build that would be dangerously clumsy.)

TBR=thestig@chromium.org (for printing)

Change-Id: I2225ded078a14b2ffcf6643e483ff995c48cb66a
Reviewed-on: https://chromium-review.googlesource.com/947398
Commit-Queue: Nick Carter <nick@chromium.org>
Reviewed-by: Charlie Reis <creis@chromium.org>
Reviewed-by: Alex Moshchuk <alexmos@chromium.org>
Cr-Commit-Position: refs/heads/master@{#542288}

I'm ok for M66 merge based on comment #4.
+camsso@, are you ok with this merge to M66?

All good!

Thanks!  Merge on the way.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/eaf1a65d2467fb665ddc6848eab79a6400b865e4

commit eaf1a65d2467fb665ddc6848eab79a6400b865e4
Author: Charlie Reis <creis@chromium.org>
Date: Mon Mar 12 17:06:41 2018

Merge to M66: Add a flag that has the effect of force-disabling the site isolation field trial.

It's a flag, rather than a Feature, so that it can be marked as unsafe later on.

(Ideally we would make the existing #enable-site-per-process flag multi-state,
but there appears to be no ready way to do that without losing the old setting:
there's no migration scheme at present, and since flags storage isn't
versioned, all the ways I can see to build that would be dangerously clumsy.)

TBR=nick@chromium.org, thestig@chromium.org (for printing)
BUG= 817137 

(cherry picked from commit 855bc498e787bb5cbbb393769ce2631bc42ab90d)

Change-Id: I2225ded078a14b2ffcf6643e483ff995c48cb66a
Reviewed-on: https://chromium-review.googlesource.com/947398
Commit-Queue: Nick Carter <nick@chromium.org>
Reviewed-by: Charlie Reis <creis@chromium.org>
Reviewed-by: Alex Moshchuk <alexmos@chromium.org>
Cr-Original-Commit-Position: refs/heads/master@{#542288}
Reviewed-on: https://chromium-review.googlesource.com/959225
Cr-Commit-Position: refs/branch-heads/3359@{#165}
Cr-Branched-From: 66afc5e5d10127546cc4b98b9117aff588b5e66b-refs/heads/master@{#540276}
[modify] https://crrev.com/eaf1a65d2467fb665ddc6848eab79a6400b865e4/chrome/browser/about_flags.cc
[modify] https://crrev.com/eaf1a65d2467fb665ddc6848eab79a6400b865e4/chrome/browser/flag_descriptions.cc
[modify] https://crrev.com/eaf1a65d2467fb665ddc6848eab79a6400b865e4/chrome/browser/flag_descriptions.h
[modify] https://crrev.com/eaf1a65d2467fb665ddc6848eab79a6400b865e4/chrome/browser/policy/site_isolation_policy_browsertest.cc
[modify] https://crrev.com/eaf1a65d2467fb665ddc6848eab79a6400b865e4/components/printing/browser/print_manager_utils.cc
[modify] https://crrev.com/eaf1a65d2467fb665ddc6848eab79a6400b865e4/content/browser/isolated_origin_browsertest.cc
[modify] https://crrev.com/eaf1a65d2467fb665ddc6848eab79a6400b865e4/content/browser/site_isolation_policy.cc
[modify] https://crrev.com/eaf1a65d2467fb665ddc6848eab79a6400b865e4/content/public/common/content_switches.cc
[modify] https://crrev.com/eaf1a65d2467fb665ddc6848eab79a6400b865e4/content/public/common/content_switches.h
[modify] https://crrev.com/eaf1a65d2467fb665ddc6848eab79a6400b865e4/tools/metrics/histograms/enums.xml

Note to self - when configuring field trials, use min_version = 66.0.3359.30:
- Commit 855bc498... initially landed in 67.0.3367.0
- Merged to 66.0.3359.30 (as eaf1a65d...).