Finch cannot control experiment by channel for standalone webview, anyways. Can we just split by
standalone webview: 1%
monochrome canary/dev/beta: 100%
monochrome stable: 1% ?

which can be done without Finch anyways.

we don't build code conditioned on channel and we should not start

We can't make the .so different for different channels, but we totally do check the channel at run-time (see version_info::GetChannel). Can we not set minidump_generation_user_fraction at run-time?

Yeah for monochrome we can do that, we'd just have to fix it for the standalone webview if we went that way.

The advantage of finch is that we can change it for the chrome version, instead of for the channel, so we can make it 1% for 71 when we push 71 to stable and that will affect both standalone and monochrome..

Re #4, "when we push X to stable" this sounds like the same gnarly task.

Toby, are we ok with limiting crash rates to 1% for standalone?

I guess the question is, "what's the last time we got an important crash that only appeared in stand-alone WebView beta?" I don't know of any recent ones.

hazems@, would you be interested in taking this?

Hmm... I tried to compare the crash counts from standalone vs monochrome.

According to
https://developer.android.com/about/dashboards/

L + M = 39.2%, N + O = 49.7%

So more Monochrome installs than standalone webview installs.

However, we seem to get more crashes from empty channel than from beta channel in pre-stable stage M71:
https://crash.corp.google.com/browse?q=product_name%3D%27AndroidWebView%27+AND+product.Version%3D%2771.0.3578.64%27
(see channel: 5773 vs 1061)

I think empty channel was for standalone webview,
but SDK versions give somewhat different numbers:
(21-23 are L + M which are 51%.)

1	23	26.59%	1817
2	26	19.20%	1312
3	22	15.06%	1029
4	28	14.62%	999
5	27	10.07%	688
6	21	9.80%	670
7	24	3.26%	223
8	25	1.33%	91
9	0	0.04%	3
10	empty	0.03%	2

100
Total:	100.00%	6834

Also, the numbers are not large enough to sample at 1%, if we try to slice it down to a version and a crash signature.

Therefore, 1% sampling for standalone webview beta may significantly (50-80%) lower our incoming crash counts for pre-stable webview.

Can this be done ASAP? Less than 1 week to stable.

It sounds like we're not going to try anything special for 71, so here:
https://chromium-review.googlesource.com/c/chromium/src/+/1356222

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/88f8ad05fbfdc9d736f58b3a98cd15784e7b1ff9

commit 88f8ad05fbfdc9d736f58b3a98cd15784e7b1ff9
Author: Torne (Richard Coles) <torne@google.com>
Date: Fri Nov 30 08:18:29 2018

Lower minidump generation fraction to 1% for stable.

To avoid hitting the cap on the number of uploaded crash reports per day
we lower the fraction of uploaded reports from 100% for Beta to 1% for
Stable.

(cherry picked from commit 4976956a14ebaca0f381e8d7f6b4241290aaf19d)

BUG= 901005 

Change-Id: Ice53a40087d734b2b5b8baf9b5ba88535628bc99
Reviewed-on: https://chromium-review.googlesource.com/c/1356222
Reviewed-by: Ben Mason <benmason@chromium.org>
Reviewed-by: Tobias Sargeant <tobiasjs@chromium.org>
Cr-Commit-Position: refs/branch-heads/3578@{#858}
Cr-Branched-From: 4226ddf99103e493d7afb23a4c7902ee496108b6-refs/heads/master@{#599034}
[modify] https://crrev.com/88f8ad05fbfdc9d736f58b3a98cd15784e7b1ff9/android_webview/common/crash_reporter/aw_crash_reporter_client.cc

Here's a summary of the rules that were executed: 
 - OnlyMergeApprovedChange: Rule Failed -- Revision 88f8ad05fbfdc9d736f58b3a98cd15784e7b1ff9 was merged to refs/branch-heads/3578 branch with no merge approval from a TPM! 
Please explain why this change was merged to the branch!
 - AcknowledgeMerge: Notification Required -- 

The following revision refers to this bug: 
https://chromium.googlesource.com/chromium/src.git/+/88f8ad05fbfdc9d736f58b3a98cd15784e7b1ff9

Commit: 88f8ad05fbfdc9d736f58b3a98cd15784e7b1ff9
Author: torne@google.com
Commiter: tobiasjs@chromium.org
Date: 2018-11-30 08:18:29 +0000 UTC

Lower minidump generation fraction to 1% for stable.

To avoid hitting the cap on the number of uploaded crash reports per day
we lower the fraction of uploaded reports from 100% for Beta to 1% for
Stable.

(cherry picked from commit 4976956a14ebaca0f381e8d7f6b4241290aaf19d)

BUG= 901005 

Change-Id: Ice53a40087d734b2b5b8baf9b5ba88535628bc99
Reviewed-on: https://chromium-review.googlesource.com/c/1356222
Reviewed-by: Ben Mason <benmason@chromium.org>
Reviewed-by: Tobias Sargeant <tobiasjs@chromium.org>
Cr-Commit-Position: refs/branch-heads/3578@{#858}
Cr-Branched-From: 4226ddf99103e493d7afb23a4c7902ee496108b6-refs/heads/master@{#599034}

Minidump not uploading, verified on Pixel 3 / PQ2A.181115.003 on  build 71.0.3578.83.

Verified on 71.0.3578.83, With the flags ONLY , minidump uploading otherwise, Crash ID not generated as cant be manually verified.Report ID generated after enabling through flags : 1543862597,6fa4f2557ee8d3eb
1543862626,6549d7df346cdcfb