Closed Bug 1348127 Opened 7 years ago Closed 7 years ago

Funnelcakes 110/111/112/113 Tour Notification, Automigration

Categories

(Release Engineering :: Release Requests, defect)

Product:
Release Engineering
Component:
Release Requests
Platform:
Unspecified
Windows
Type:
defect
Priority:
Not set
Severity:
normal

Tracking

(firefox53+ fixed)

Status:
RESOLVED FIXED
Tracking Flags:
Tracking Status
firefox53 + fixed

People

(Reporter: nyee, Unassigned)

References

(Depends on 1 open bug, Blocks 1 open bug)

Details

Attachments

(5 files, 8 obsolete files)

Funnel Cake Addon.JPG
59.97 KB, image/jpeg
Details
Screen Shot 2017-04-18 at 13.44.28.png
86.19 KB, image/png
Details
Capture.JPG
70.84 KB, image/jpeg
Details
firefox_onboard-1.0.6-fx.xpi
35.56 KB, application/zip
Details
Error when uploading XPI to AMO
26.29 KB, image/png
Details 
      No description provided.
Blocks: 1348126
We are building a new version of the onboarding experience which includes a new add-on XPI.

This is similar to the approach in bug 1332726 (funnelcakes 101-104) but also includes the streamlined stub installer and automigration from funnelcakes 99/100. (bug 1333873)

Target launch date is April 20th, 2017, with retention measurements starting April 23, 2017.

Variations:
110 - Control, vanilla Firefox 53 with just the configs to denote the funnelcake
111 - Streamlined stub installer (bugs 1333873, 1350405) + Automigration (bugs 1271775, 1271799, + fixes under 1332225) + tour notifications variant A
112 - Streamlined stub installer (bugs 1333873, 1350405) + Automigration (bugs 1271775, 1271799, + fixes under 1332225) + tour notifications variant B
113 - Streamlined stub installer (bugs 1333873, 1350405) + without automigration + tour notifications variant A

Details on tour variants: https://docs.google.com/document/d/1008Na15VH7z2ZRuVihTI1abk9Fr5-K1DqSUUFsXQk9Y/edit#heading=h.rgj4t957f9n2
Summary: Creating funnelcake builds for Tour Notification → Funnelcakes 110/111/112/113 Tour Notification, Automigration
Depends on: 1350974
Depends on: 1350405
Could you let me know what Firefox pref to set to control tour notifications A vs B ?
Flags: needinfo?(nyee)
Not sure this is the right pages but the first run pages need to pass tracking parameters to FxA on a different way. 

Currently you are passing funnel cake versions via entrypoint. For the time being until we have designated parameters for this, utm_source would be best.

Entrypoint is used by engineering to customize the flow so it could have negative repercussions. 

There are 2 ways to make this change:
1) in the first run pages. If you have different first run pages per funnel cake, you can specify the right parameters in the page
2) In a less clean manner, we could change the FxA iframe parameters with optimizely based the url of the first run pages that has funnel cake numbers. While this works, it's a bit hacky and would require an optimizely test for each funnel cake iiuc.
Flags: needinfo?(pdolanjski)
Oh, this also includes the control. Just thought I'd specify.
(In reply to Nick Thomas [:nthomas] from comment #2)
> Could you let me know what Firefox pref to set to control tour notifications
> A vs B ?

Peter would have the correct answer for you.
Flags: needinfo?(nyee)
Here is what I need to know by the end of the week to write the configuration for the full installer builds:
* confirmation that we're targeting en-US for 32bit Windows, or if there are more locales or win64 too
* the set of 4 firstrun urls we are going to use (previously we've set startup.homepage_welcome_url="https://www.mozilla.org/%LOCALE%/%APP%/%VERSION%/firstrun/?f=NNN")
* the preference to set to control tour A vs tour B
We also need the following preferences set:

distribution.variation, this will be set as either contentVariationA or contentVariationB
browser.newtab.preload, false
browser.newtab.url, about:newtab

Thanks!
(In reply to Alex Davis [:adavis] [PM FxA+Sync] from comment #3)
> Not sure this is the right pages but the first run pages need to pass
> tracking parameters to FxA on a different way. 
> 
> Currently you are passing funnel cake versions via entrypoint. For the time
> being until we have designated parameters for this, utm_source would be best.
> 
> Entrypoint is used by engineering to customize the flow so it could have
> negative repercussions. 
> 
> There are 2 ways to make this change:
> 1) in the first run pages. If you have different first run pages per funnel
> cake, you can specify the right parameters in the page
> 2) In a less clean manner, we could change the FxA iframe parameters with
> optimizely based the url of the first run pages that has funnel cake
> numbers. While this works, it's a bit hacky and would require an optimizely
> test for each funnel cake iiuc.

Cmore, any suggestions on how to proceed with this?
Flags: needinfo?(chrismore.bugzilla)
Also creating a separate bug for this so we can track it.
Depends on: 1354710
The refreshed and fixed streamlined stub is ready for testing, see bug 1353576. The small patches needed for the stub for each particular funnelcake version/id are also there. Let me know if any more is needed from the stub side.
Thanks Adam, any blockers to me going ahead and generating stubs on mozilla-release ?

Moving this bug to the usual component for RelEng's part of funnelcake. Johan Lorenzo (jlorenzo) and Aki Sasaki (aki) are on releaseduty for Firefox 53.0, so they're on the hook. I'm around another 24 hours before starting PTO, getting things set up as much as possible and documented for them.
Component: General → Releases: Custom Builds
Product: Firefox → Release Engineering
QA Contact: coop
Version: 52 Branch → unspecified
There are no blockers I know of, I've done a small amount of local testing but this is low risk as it is very similar to the last time. I don't see a good reason not to build the stubs now, might as well do testing on the exact code that will be going out.

So please go ahead and build the stubs, thanks Nick!
Stub generation underway, see bug 1353576 comment #17.
Assuming Nicole has this under control but I'll track it for 53 so that I can be aware of the progress after we release.
status-firefox53: --- → affected
tracking-firefox53: --- → +
Clearing the ni requests because they've been resolved in bug 1354710.

We'll be using these bouncer URLs (extrapolating from previous funnelcakes, in particular bug 1326252), first the stubs:
https://download.mozilla.org/?product=firefox-stub-f110&os=win&lang=en-US
https://download.mozilla.org/?product=firefox-stub-f111&os=win&lang=en-US
https://download.mozilla.org/?product=firefox-stub-f112&os=win&lang=en-US
https://download.mozilla.org/?product=firefox-stub-f113&os=win&lang=en-US

And the full installers which the stub will grab:
https://download.mozilla.org/?product=firefox-latest-f110&os=win&lang=en-US
https://download.mozilla.org/?product=firefox-latest-f111&os=win&lang=en-US
https://download.mozilla.org/?product=firefox-latest-f112&os=win&lang=en-US
https://download.mozilla.org/?product=firefox-latest-f113&os=win&lang=en-US

The first set of links work (after bug 1353576 comment #18), but the second set won't until we generate some builds here. We're waiting for the tour notification addon before that can happen. There is a pull request at https://github.com/mozilla-partners/funnelcake/pull/53 which contains the rest of the config; the addon can be added to that.

I'm now handing off work to jlorenzo and aki, the releaseduty people for 53.0. They have lots of docs out of band, and jlund to ask for advice.
Flags: needinfo?(pdolanjski)
Flags: needinfo?(chrismore.bugzilla)
Attached file @onboard-v1-1.0.0.xpi self-signed XPI (obsolete) — 
I pulled down the branch of funnelcake-builds from https://github.com/mozilla/onboard/tree/funnelcake-builds

Ran sudo jpm xpi

and it created the XPI that is attached. This should be good enough to create the first funnelcake with the XPI bundled for testing.

We will need to submit another later to AMO after a funnelcake+XPI has been tested and verified to be doing the right things.
Looks like we're ready for some SV QA. Justin can you help out here?
Flags: needinfo?(jwilliams)
Yes I will get on this. Thanks Nicole.
Flags: needinfo?(jwilliams)
Attached image Funnel Cake Addon.JPG — 
There is an issue when adding the signed xpi to Nightly.
Flags: needinfo?(chrismore.bugzilla)
please see comment 19
Flags: needinfo?(nyee)
Cmore, there is a newer version on GitHub, if you can use that one to try to resolve the above corruption issue.
Attached file @onboard-v1-1.0.0.xpi UN signed XPI (obsolete) — 
ran: 

sudo git pull
sudo jpm xpi

Created the attached XPI.

Should I sign the XPI too? This is an unsigned XPI.
Attachment #8858165 - Attachment is obsolete: true
Flags: needinfo?(chrismore.bugzilla)
I get the same corruption error with the new xpi provided in comment 22.
(In reply to Justin [:JW_SoftvisionQA] from comment #23)
> I get the same corruption error with the new xpi provided in comment 22.
Flags: needinfo?(nyee) → needinfo?(chrismore.bugzilla)
I tried to run jpm xpi sign --api-key user: XXXXXXX

and got this error:

JPM [info] Created XPI for signing: /tmp/tmp-unsigned-xpi-1026613zUfL82qiJk/@onboard-v1-1.0.0.xpi
JPM [error] Server response: No install.rdf or manifest.json found ( status: 400 )
JPM [info] FAIL
Flags: needinfo?(chrismore.bugzilla) → needinfo?(schalk.neethling.bugs)
Note, the XPI in comment 16 was not self-signed. I just ran jpm xpi to build the XPI. Comment 25 was my first attempt at doing self-signing and thus I got the manifest.json error.
Like the error message indicates, the package doesn't have an install.rdf, so something went wrong when creating the XPI file. Were there no error messages when creating it?
Flags: needinfo?(chrismore.bugzilla)
(In reply to Jorge Villalobos [:jorgev] from comment #27)
> Like the error message indicates, the package doesn't have an install.rdf,
> so something went wrong when creating the XPI file. Were there no error
> messages when creating it?

Manual XPI creation:

Chriss-MacBook-Air:onboard cmore$ sudo jpm xpi
JPM [info] Starting jpm xpi on Firefox OnBoard
JPM [info] Successfully created XPI at /Users/cmore/onboard/@onboard-v1-1.0.0.xpi (131ms)

No error when creating the XPI manually.

I got the error in comment 25 when trying to xpi sign the add-on.
Flags: needinfo?(chrismore.bugzilla)
Attached file onboard-v1.xpi (obsolete) — 
I did the following (on OS X, fwiw):

1) Downloaded the XPI from comment #22
2) Unpacked it.
3) Ran `jpm xpi` on the unpacked directory.

I got the attached file as a result, which does have an install.rdf. I haven't tested if it works correctly.
(In reply to Jorge Villalobos [:jorgev] from comment #29)
> Created attachment 8858906 [details]
> onboard-v1.xpi
> 
> I did the following (on OS X, fwiw):
> 
> 1) Downloaded the XPI from comment #22
> 2) Unpacked it.
> 3) Ran `jpm xpi` on the unpacked directory.
> 
> I got the attached file as a result, which does have an install.rdf. I
> haven't tested if it works correctly.

Justin, Is this xpi able to work for you?
Flags: needinfo?(jwilliams)
Justin will be back a little later, so I am answering in his stead. I was able to install the xpi with no corruption issues with the attachment provided in Comment 29.
Flags: needinfo?(jwilliams)
(In reply to Grover Wimberly IV [:Grover-QA] from comment #31)
> Justin will be back a little later, so I am answering in his stead. I was
> able to install the xpi with no corruption issues with the attachment
> provided in Comment 29.

Great. Thanks Grover for responding. 

Johan, Jordan and Aki - We are able to install the xpi with no corruption. I am giving the green light to start building the staging funnelcake builds using this xpi. assuming its self-signed.
Flags: needinfo?(jlund)
Flags: needinfo?(jlorenzo)
Flags: needinfo?(aki)
Justin or Grover: Here is our test plan: https://docs.google.com/a/mozilla.com/document/d/1i5FQk0CctFnD9xGDaLdiAkTyL-r2pBg_lScJlO1k3aU/edit?usp=sharing

Once the funnelcakes are built using the self-signed xpi, we can begin testing.
Flags: needinfo?(jwilliams)
Flags: needinfo?(gwimberly)
Flags: needinfo?(gwimberly)
Starting the funnelcake configs per https://gist.github.com/nthomas-mozilla/feee7a521c1472a9a8483f734f18e96e#--generate-full-installer-configs
Flags: needinfo?(jlund)
Flags: needinfo?(jlorenzo)
Flags: needinfo?(aki)
(In reply to Nicole Yee (:nicoleyee) from comment #32)
> assuming its self-signed.

The XPI in comment #29 isn't signed.
(In reply to Jorge Villalobos [:jorgev] from comment #35)
> (In reply to Nicole Yee (:nicoleyee) from comment #32)
> > assuming its self-signed.
> 
> The XPI in comment #29 isn't signed.

Cmore, can you help get this signed for us.
Flags: needinfo?(chrismore.bugzilla)
(In reply to Nicole Yee (:nicoleyee) from comment #36)
> (In reply to Jorge Villalobos [:jorgev] from comment #35)
> > (In reply to Nicole Yee (:nicoleyee) from comment #32)
> > > assuming its self-signed.
> > 
> > The XPI in comment #29 isn't signed.
> 
> Cmore, can you help get this signed for us.

I can't. As mentioned in comment 25, I get an error when I try to self-sign the XPI. I haven't got these errors before when signing.

I can try uploading it to AMO and go through that process, but I am worried it will be kicked back there. The error in comment 25 is something new that I haven't got the Schalk's previous onboarding add-ons.

Can anyone else from the AMO team or someone else with technical skills to take over this part? There's nothing special here I am doing other than running jpm locally on my mac and producing an XPI. Then after that, I would just upload the XPI to AMO for review/final signing.
Flags: needinfo?(chrismore.bugzilla) → needinfo?(nyee)
I now have access to the host partner-repack1.  However, should we hold off on building this until we get the xpi signed?
I sent a mail out to some people on this bug but, adding here for those that were not and posterity 

"Ok so here is what I believe is happening. Yesterday I noticed that the notice about the end of SDK add-ons have been updated on MDN and now also reads:

> From Firefox 53 onwards, no new legacy add-ons will be accepted on addons.mozilla.org (AMO).
> From Firefox 57 onwards, WebExtensions will be the only supported extension type, and Firefox will not load other types.

From the errors I can see that AMO has made the switchover and are no longer accepting SDK based add-ons hence the errors that are being seen. Firefox 53 is being released to day no?

This presents a couple of problems:

1. Self sign will not work(as we currently see) and this will prevent Funnelcake builds
2. When we want to create a final build, we need to upload the XPI to AMO and, here it will be kicked back, and I do not see how they will make an exception for one specific add-on(there might be some internal ways around this but, not sure.)

Here is my proposal for a possible solution. We repackage(there will be some development needed) the add-on as a "transitional" add-on with an empty web-extension in it's belly. This will mean that the file it is looking for will be present and it will pass AMO and be installable post Fx 53.

I will create a separate branch into which I will stick this code and will ping this list once complete."
Flags: needinfo?(schalk.neethling.bugs)
Shows screenshot of extracted XPI with install.rdf as well as manifest.json
Here is the PR that turns this into a embedded web extension:
https://github.com/mozilla/onboard/pull/108

This now has an `install.rdf` file as well as a `manifest.json` - See the attached file(https://bugzilla.mozilla.org/show_bug.cgi?id=1348127#c40).

:cmore, please sign this and let me know if this works out. The branch to use is:
https://github.com/mozilla/onboard/tree/onboard-embedded-webext
Flags: needinfo?(chrismore.bugzilla)
I got the same error. Let me put in exactly what I am doing:

1) sudo git clone https://github.com/mozilla/onboard.git

2) cd onboard

4) sudo git checkout onboard-embedded-webext

4) sudo jpm sign --api-key user:12393155:473 --api-secret XXXXXX

Error:

JPM [info] Created XPI for signing: /tmp/tmp-unsigned-xpi-11182fcyNGOVm3bii/@onboard-v1-1.0.0.xpi
JPM [error] Server response: No install.rdf or manifest.json found ( status: 400 )
JPM [info] FAIL

I even tried:

git pull origin onboard-embedded-webext 

to confirm the local branch is up-to-date. Same error.

:(

Am I doing something wrong??
Flags: needinfo?(chrismore.bugzilla)
* numbering should be 1, 2, 3, 4
:cmore, we will need to follow up with AMO on this as the XPI that is created definitely contains both of those entries. Which version of jpm are you using? Should be 1.2.2
Flags: needinfo?(chrismore.bugzilla)
Attached file firefox_onboard-1.0.0-fx.xpi (obsolete) — 
Latest signed XPI
Flags: needinfo?(chrismore.bugzilla)
Attached file firefox_onboard-1.0.1-fx.xpi (obsolete) — 
Latest signed XPI built from master.
Attachment #8858869 - Attachment is obsolete: true
Attachment #8858906 - Attachment is obsolete: true
Attachment #8859249 - Attachment is obsolete: true
(In reply to Schalk Neethling [:espressive] from comment #46)
> Created attachment 8859250 [details]
> firefox_onboard-1.0.1-fx.xpi
> 
> Latest signed XPI built from master.

ready!
Flags: needinfo?(nyee) → needinfo?(aki)
Looks like it was my local jpm version being too old (1.0.7). Now we know, but Schalk is able to sign now. win win!
Thanks! On it.
Flags: needinfo?(aki)
Ok so, just to note, this funnelcake build is for testing so the timers are set at 2 minutes. The next build will have the full 24 hour timers.
I just tested this with the addon and everything looks fine. Waiting on the Funnelcake builds to test and signoff. Once we get the build if everything works as expected testing should only take 4 hours or so.
Flags: needinfo?(jwilliams) → needinfo?(nyee)
Thanks Justin! Since the addon looks good, should we submit it to AMO? Or should we test with the build first, then submit it to AMO?
Flags: needinfo?(nyee) → needinfo?(chrismore.bugzilla)
(In reply to Nicole Yee (:nicoleyee) from comment #52)
> Thanks Justin! Since the addon looks good, should we submit it to AMO? Or
> should we test with the build first, then submit it to AMO?

I would say let's first test the build. I will need to build another XPI to submit to AMO that does not use the testing timer duration of 2 minutes.
(In reply to Schalk Neethling [:espressive] from comment #53)
> (In reply to Nicole Yee (:nicoleyee) from comment #52)
> > Thanks Justin! Since the addon looks good, should we submit it to AMO? Or
> > should we test with the build first, then submit it to AMO?
> 
> I would say let's first test the build. I will need to build another XPI to
> submit to AMO that does not use the testing timer duration of 2 minutes.

Yeah, let's wait for the first build to make sure it all functions correctly so to not have to go through the AMO queue more than once.
Flags: needinfo?(chrismore.bugzilla)
https://archive.mozilla.org/pub/firefox/candidates/53.0-candidates/build6/partner-repacks/funnelcake110/v1/win32/en-US/Firefox%20Setup%2053.0.exe
SHA256(repacked_builds/signed/partner-repacks/funnelcake110/v1/win32/en-US/Firefox Setup 53.0.exe)= 2c6d7bc39bc8aa075d42a63204e50c40fdb6cac4de402de1cc384f116f239724

https://archive.mozilla.org/pub/firefox/candidates/53.0-candidates/build6/partner-repacks/funnelcake111/v1/win32/en-US/Firefox%20Setup%2053.0.exe
SHA256(repacked_builds/signed/partner-repacks/funnelcake111/v1/win32/en-US/Firefox Setup 53.0.exe)= b03b430ce0fce2ad99202372bfd3bf31990f4efaef02197188566f16ac7f26de

https://archive.mozilla.org/pub/firefox/candidates/53.0-candidates/build6/partner-repacks/funnelcake112/v1/win32/en-US/Firefox%20Setup%2053.0.exe
SHA256(repacked_builds/signed/partner-repacks/funnelcake112/v1/win32/en-US/Firefox Setup 53.0.exe)= 0f0d28388b764633d3fdb8e009726a2080d983672c3c517bf9585eb7128080ba

https://archive.mozilla.org/pub/firefox/candidates/53.0-candidates/build6/partner-repacks/funnelcake113/v1/win32/en-US/Firefox%20Setup%2053.0.exe
SHA256(repacked_builds/signed/partner-repacks/funnelcake113/v1/win32/en-US/Firefox Setup 53.0.exe)= b2b4a1cb00836730f5db2ab8b309bcbe0330b87e8fdc56e21f3a73b521c183bf
Hey Schalk,

We'll want to set up separate data sets for our different environments (dev, staging, prod - Or whatever is relevant in our case).

Can you let me know:
1. Which environments we have
2. If there will be any data point passed to GA that I can use to filter for these environments - If there is no parameter that I can use, we may need to add something.
Flags: needinfo?(schalk.neethling.bugs)
builds are up! Ready for testing.
Flags: needinfo?(jwilliams)
Alright. I will get on that.
Flags: needinfo?(jwilliams)
Attached image Capture.JPG — 
It looks like the addon is not being installed within these builds. Maybe the unsigned xpi was used to build these? The attached photo is what is showing in Funnelcake Builds 111-113. Funnelcake Build does not show the extension at all (pretty sure this is expected).
Flags: needinfo?(nyee)
Flags: needinfo?(chrismore.bugzilla)
Flags: needinfo?(aki)
(In reply to Justin [:JW_SoftvisionQA] from comment #59)
> Created attachment 8859358 [details]
> Capture.JPG
> 
> It looks like the addon is not being installed within these builds. Maybe
> the unsigned xpi was used to build these? The attached photo is what is
> showing in Funnelcake Builds 111-113. Funnelcake Build does not show the

showing in Funnelcake Builds 111-113. Funnelcake Build 110* does not show the

> extension at all (pretty sure this is expected).
hmmm. The disabled add-on usually happens when the add-on is not signed, but comment 46 has a self-signed XPI.

:aki: did you use the XPI from comment 46 in those builds?
Flags: needinfo?(chrismore.bugzilla)
I did wget -O foo.xpi https://bugzilla.mozilla.org/attachment.cgi?id=8859250 and followed the rest of the instructions in https://gist.github.com/escapewindow/5dba93a02bb346f5ac3677477106a46f#x-generate-full-installer-configs .  Do those change when we're pretending to be a web extension?

I checked in the above in https://github.com/mozilla-partners/funnelcake/commit/c3003910e948a15e357be25d2d4ff919dde8324f .
Flags: needinfo?(aki)
There were a few data points missing from what I can see in GA. (Schalk, I've pinged you in the doc at all the relevant places)

1) Add requests for launch of the app
2) Load of first run
3) Confirm if we need to re-instrument accounts or if we are using an existing version
4) Adding prefix 'step' to event actions prior to the number
5) Passing the variation into CD3

I'll ping you tomorrow morning to try to resolve any outstanding issues.
(In reply to pgerman from comment #63)
> There were a few data points missing from what I can see in GA. (Schalk,
> I've pinged you in the doc at all the relevant places)
> 
> 1) Add requests for launch of the app
> 2) Load of first run
> 3) Confirm if we need to re-instrument accounts or if we are using an
> existing version
> 4) Adding prefix 'step' to event actions prior to the number
> 5) Passing the variation into CD3
> 
> I'll ping you tomorrow morning to try to resolve any outstanding issues.

Thanks for the info. Some of these does not apply to the add-on here. I made relevant comments in the doc.

> 4) Adding prefix 'step' to event actions prior to the number
> 5) Passing the variation into CD3

These two has been resolved in the latest XPI attached to this bug.
Flags: needinfo?(schalk.neethling.bugs)
Attached file firefox_onboard-1.0.3-fx.xpi (obsolete) — 
Latest signed XPI build of the add-on.
Attachment #8859250 - Attachment is obsolete: true
Latest XPI has been uploaded: https://bugzilla.mozilla.org/show_bug.cgi?id=1348127#c65

Please NOTE the following preference we need set:

distribution.variation, this will be set as either contentVariationA or contentVariationB dependant on funnelcake build
browser.newtab.preload, false
browser.newtab.url, about:newtab

Thanks!
I generated the installer configs[1], by following [2]. I made sure the distribution variation are set as desscribed in comment 66.

However, I don't have the rights to push to the master branch of mozilla-partners/funnelcake. Can anybody merge [1] for me?

[1] https://github.com/mozilla-partners/funnelcake/pull/55
[2] https://gist.github.com/escapewindow/5dba93a02bb346f5ac3677477106a46f#x-generate-full-installer-configs
(In reply to Johan Lorenzo [:jlorenzo] from comment #67)
> However, I don't have the rights to push to the master branch of
> mozilla-partners/funnelcake.

Who does?  Is that someone on rel-eng?
(In reply to Peter Dolanjski [:pdol] from comment #68)
> (In reply to Johan Lorenzo [:jlorenzo] from comment #67)
> > However, I don't have the rights to push to the master branch of
> > mozilla-partners/funnelcake.
> 
> Who does?  Is that someone on rel-eng?

Do we need it pushed anywhere at this time? Should we not just have access to test?

I just want to be sure we do not push this somewhere where it might end up in the wild before we intend to release it.

Without Optimizely directing download traffic it most likely will not happen but, I just want to make sure we are aware that this is not yet the version to be pushed out to our target audience. Or is it?
(In reply to Justin [:JW_SoftvisionQA] from comment #60)
> (In reply to Justin [:JW_SoftvisionQA] from comment #59)
> > Created attachment 8859358 [details]
> > Capture.JPG
> > 
> > It looks like the addon is not being installed within these builds. Maybe
> > the unsigned xpi was used to build these? The attached photo is what is
> > showing in Funnelcake Builds 111-113. Funnelcake Build does not show the
> 
> showing in Funnelcake Builds 111-113. Funnelcake Build 110* does not show the
> 
> > extension at all (pretty sure this is expected).

Justin are you still blocked by this? Did the newest signed RC resolve this?
Flags: needinfo?(nyee) → needinfo?(jwilliams)
Im sorry Nicole I do not see the newest Funnelcake builds.
Flags: needinfo?(jwilliams) → needinfo?(nyee)
(In reply to Johan Lorenzo [:jlorenzo] from comment #67)
> I generated the installer configs[1], by following [2]. I made sure the
> distribution variation are set as desscribed in comment 66.
> 
> However, I don't have the rights to push to the master branch of
> mozilla-partners/funnelcake. Can anybody merge [1] for me?
> 
> [1] https://github.com/mozilla-partners/funnelcake/pull/55
> [2]
> https://gist.github.com/escapewindow/5dba93a02bb346f5ac3677477106a46f#x-
> generate-full-installer-configs

Merged.
Added a "Before you start" to my fork of nthomas' gist: https://gist.github.com/escapewindow/5dba93a02bb346f5ac3677477106a46f#before-you-start
(In reply to Justin [:JW_SoftvisionQA] from comment #60)
> (In reply to Justin [:JW_SoftvisionQA] from comment #59)
> > Created attachment 8859358 [details]
> > Capture.JPG
> > 
> > It looks like the addon is not being installed within these builds. Maybe
> > the unsigned xpi was used to build these? The attached photo is what is
> > showing in Funnelcake Builds 111-113. Funnelcake Build does not show the
> 
> showing in Funnelcake Builds 111-113. Funnelcake Build 110* does not show the
> 
> > extension at all (pretty sure this is expected).

110 is the control (just current release version), so that makes sense.
Yes, I am currently blocked as I wait for Funnelcake 111-113. I can not signoff on just 110 since it does not include the Funnelcake changes.
I have to create the funnelcakes now that the new apk is merged (comment 110); currently I'm dealing with some release bustage, but this funnelcake is next on my todo list.
(In reply to Aki Sasaki [:aki] from comment #75)
> I have to create the funnelcakes now that the new apk is merged (comment
> 110); currently I'm dealing with some release bustage, but this funnelcake
> is next on my todo list.

s,apk,xpi,
https://archive.mozilla.org/pub/firefox/candidates/53.0-candidates/build6/partner-repacks/funnelcake110/v2/win32/en-US/Firefox%20Setup%2053.0.exe
SHA256(repacked_builds/signed/partner-repacks/funnelcake110/v2/win32/en-US/Firefox Setup 53.0.exe)= 86f6169a36915fa3130f6899c2b1bdcb529356779e82dc7ba7e20fcfabd18910

https://archive.mozilla.org/pub/firefox/candidates/53.0-candidates/build6/partner-repacks/funnelcake111/v2/win32/en-US/Firefox%20Setup%2053.0.exe
SHA256(repacked_builds/signed/partner-repacks/funnelcake111/v2/win32/en-US/Firefox Setup 53.0.exe)= bf57bef1c3f559b13caac1205326b89fb93c80e69d4ba7934dd07ec41fc52151

https://archive.mozilla.org/pub/firefox/candidates/53.0-candidates/build6/partner-repacks/funnelcake112/v2/win32/en-US/Firefox%20Setup%2053.0.exe
SHA256(repacked_builds/signed/partner-repacks/funnelcake112/v2/win32/en-US/Firefox Setup 53.0.exe)= ad0f4e8aeb6f4b8d2f1ab8cc903ab8c4c619f8cccd89fb143ad641dd479d46e4

https://archive.mozilla.org/pub/firefox/candidates/53.0-candidates/build6/partner-repacks/funnelcake113/v2/win32/en-US/Firefox%20Setup%2053.0.exe
SHA256(repacked_builds/signed/partner-repacks/funnelcake113/v2/win32/en-US/Firefox Setup 53.0.exe)= 928e54a958b95b2af6c132cb7d9765e7291979fcadd945bcf5ceb8d970396347
I can verify that the addon is being installed on Funnelcake 111. I will start testing this now. We should finish testing this some time tomorrow. Thank you everyone that helped get this in motion!
Is the time on these builds set to 2 minutes or 24 hours? I've looked at Funnelcake 111 and 112 and have not received the first notification.
Flags: needinfo?(pdolanjski)
Yes, it seems like the timer is set to 24 hours. That is not a problem as I can just forward my system clock a couple of days to speed up the process.
(In reply to Justin [:JW_SoftvisionQA] from comment #80)
> Yes, it seems like the timer is set to 24 hours. That is not a problem as I
> can just forward my system clock a couple of days to speed up the process.

Yup, that should be right.  We just advance the system clock as you describe.
Flags: needinfo?(pdolanjski)
Just an FYI and a "anyone else seeing this?"

So, I have done some testing with the funnelcakes but I am running into this thing where Windows resets the time to quickly and then the experience does not move forward. I did unset the option to "set time automatically" but, even though it allows me to then change the date and time, the above still happens. Perhaps it is because it is a VM, not sure. Anyone else seeing this? Any other places in Windows 10 where one needs to change a setting?

I will attempt with a Windows 7 VM and see if the problem happens there as well.
For one thing, you probably want to disable time sync in the VM, here is how to do it with VirtualBox:
http://rickguyer.com/virtualbox-disable-time-sync-between-host-and-client/
This is an issue we ran into here a while back with our virtual machines. For this to work, you have to change the system clock of your actual machine. The virtual box will mirror the changes.
Flags: needinfo?(schalk.neethling.bugs)
One minor issue that I did come across.

On Funnelcake 112, The notification for Sync does not take you to the sync page. You are redirected to a blank screen.
Flags: needinfo?(pdolanjski)
(In reply to Justin [:JW_SoftvisionQA] from comment #85)
> One minor issue that I did come across.
> 
> On Funnelcake 112, The notification for Sync does not take you to the sync
> page. You are redirected to a blank screen.
Flags: needinfo?(nyee)
(In reply to Justin [:JW_SoftvisionQA] from comment #85)
> One minor issue that I did come across.
> 
> On Funnelcake 112, The notification for Sync does not take you to the sync
> page. You are redirected to a blank screen.

:JW_SoftvisionQA Is there a URL for the blank page? Or is it simply a blank page with nothing in the URL bar?
Flags: needinfo?(schalk.neethling.bugs) → needinfo?(jwilliams)
Flags: needinfo?(pdolanjski)
Yes the url is:

about:accounts?action=signup&entrypoint=uitour&utm_source=onboarding-addon&utm_medium=referral&utm-campaign=contentVariationB
Flags: needinfo?(jwilliams) → needinfo?(schalk.neethling.bugs)
We have completed Testing for this feature on Win 7, 8.1, and 10. One minor issue was found. Please see comment 85 for info.

Also the Stub Installer was not tested. Please see https://bugzilla.mozilla.org/show_bug.cgi?id=1353576#c22

Please let me know if there is anything else that needs to be done on my end.
Flags: needinfo?(pdolanjski)
Flags: needinfo?(nyee)
(In reply to Justin [:JW_SoftvisionQA] from comment #88)
> Yes the url is:
> 
> about:accounts?action=signup&entrypoint=uitour&utm_source=onboarding-
> addon&utm_medium=referral&utm-campaign=contentVariationB

That URL is exactly as it should be so, there might have been an temporary problem on Fx Accounts side. Can you replicate it consistently? Perhaps there is a config issue?
Flags: needinfo?(schalk.neethling.bugs)
Yes I can replicate it across all 3 platforms. The page just shows as an all white blank page. see video below.

https://jwilliams-softvision.tinytake.com/sf/MTUzMTE2OF81MzEzMTE2
(In reply to Justin [:JW_SoftvisionQA] from comment #91)
> Yes I can replicate it across all 3 platforms. The page just shows as an all
> white blank page. see video below.
> 
> https://jwilliams-softvision.tinytake.com/sf/MTUzMTE2OF81MzEzMTE2

Adavis, any ideas why this isn't working?
Flags: needinfo?(pdolanjski) → needinfo?(adavis)
> On Funnelcake 112, The notification for Sync does not take you to the sync page. You are redirected to a blank screen.
> about:accounts?action=signup&entrypoint=uitour&utm_source=onboarding-addon&utm_medium=referral&utm-campaign=contentVariationB

Justin, can I get some more context? What notification? When do you see this?

ni? Rfk for context and in case he knows what the problem is here.
Flags: needinfo?(rfkelly)
Flags: needinfo?(jwilliams)
Flags: needinfo?(adavis)
Hi Alex. The notification can be seen in the video I provided in comment 91. I wouldn't know much more beyond that.
Flags: needinfo?(jwilliams) → needinfo?(adavis)
> utm-campaign=contentVariationB

BTW is this a config typo somewhere, IIUC this should be "utm_campaign" with an underscore.
> Rfk for context and in case he knows what the problem is here.

Unfortunately I was not able to reproduce; it would be worth taking a look in the devtools console to see if any JS errors are being logged when it fails to load the page.
Flags: needinfo?(rfkelly)
(In reply to Ryan Kelly [:rfkelly] from comment #95)
> > utm-campaign=contentVariationB
> 
> BTW is this a config typo somewhere, IIUC this should be "utm_campaign" with
> an underscore.

I think this is in the addon.
https://github.com/mozilla-partners/funnelcake/blob/master/desktop/funnelcake112/distribution/extensions/%40onboard-v1/locale/en-US.properties
(In reply to Ryan Kelly [:rfkelly] from comment #95)
> > utm-campaign=contentVariationB
> 
> BTW is this a config typo somewhere, IIUC this should be "utm_campaign" with
> an underscore.

Yes, This is a typo. You are correct. This should be fixed.

Unfortunately though, I don't think it is the cause of the problem. (which I can't seem to replicate)
Flags: needinfo?(adavis)
Attached file firefox_onboard-1.0.4-fx.xpi (obsolete) — 
Latest XPI with atm parameter typo fix
Attachment #8859488 - Attachment is obsolete: true
Hey All,

Uploaded the latest XPI with the typo fix.

Also please note, during internal testing it has been noticed that these funnelcakes don’t seem to use the v2.0 stub installer, and they have a ‘create an account’ page, not a sign in page on first run. These builds should have the Tour Notifications as well as the custom download page, stub installer, and first run page from v2.0

Also, it does not seem that these builds include auto migration?

I would also ask that we no longer set `browser.newtab.url: about:newtab` as this has no discernable effect but instead, leave it at it's default value. Which I believe is an empty string.

I am ni :pdol who can provide additional information an access to any other configuration that is needed to enable these.

Thanks so much!
Flags: needinfo?(pdolanjski)
I am not able to get any of the notifications with the new xpi from comment 99.

I can confirm that automigration is indeed working. Once I open a new tab all of my data that was imported over from chrome is shown.
(In reply to Justin [:JW_SoftvisionQA] from comment #101)
> I am not able to get any of the notifications with the new xpi from comment
> 99.

Are you 100% sure you cleaned out all previous data from Local, LocalLow, and Roaming in Users/username/AppData? Nothing has changed with the code other than the typo with regards to the utm parameter.
The xpi is to be installed on the current nightly build for testing purposes, correct?
(In reply to Justin [:JW_SoftvisionQA] from comment #103)
> The xpi is to be installed on the current nightly build for testing
> purposes, correct?

Just thought about it, if it is not build into a funnelcake it will not work as some preferences it depends on will not be set(this XPI does not fake the params). So, I would just use the existing funnelcakes for testing(with the caveat that the typo in the about:accounts url still exists in them) until a new set of funnelcakes has been spun up.
Got thank you.
(In reply to Schalk Neethling [:espressive] from comment #100)
> it has been noticed that these funnelcakes don’t seem to use the v2.0 stub installer

The funnelcake repacks in this thread are just the full installer, which will be downloaded and run silently by the stub. (Unless you're seeing the original stub installer somewhere?)

I assume by v2.0 stub installer you mean the "streamlined" stub in bug 1353576 (specified more fully for fc 99/100 in bug 1328445)? There's a signed build of that available in bug 1353576 comment 11 for testing, but note that it installs the standard release.

I've linked the proper stub builds that will use the funnelcake installers in bug 1353576 comment 22, but they will not run successfully until bouncer is serving up the repacks under the new funnelcake ids.
Schalk, have you implemented the cid that is specific to a user? (I've pinged you in the doc with the location).

Right now I'm seeing zero retention (and zero users): https://pageshot.net/35j5TfbQoGaUXzYa/analytics.google.com

Also, I've added a data points for
1) When the browser launches
2) A ping after the browser has been open for 24 hours 

https://docs.google.com/document/d/1ndh-ZGCpDZpZ4uVxnK7WftOsxjWdD_dVrxd0G9RL06c
Flags: needinfo?(schalk.neethling.bugs)
Attached file firefox_onboard-1.0.5-fx.xpi (obsolete) — 
Latest XPI. Please use this for all future funnelcake builds.

-- Adds a GA ping on each browser startup.
Attachment #8860289 - Attachment is obsolete: true
Flags: needinfo?(schalk.neethling.bugs)
Hey all,

Can we have a preference that will only be set for the final production funnelcake build? i.e. not set while the funnelcakes are still being internally tested and QA'd

If so, we need preferences set as follows:

For all funnelcakes

distribution.variation, this will be set as either contentVariationA or contentVariationB
browser.newtab.preload, false

For the production build, we also need

onboard.environment, 'production' // this is needed for Google Analytics

Let me know, thanks!
After discussing with Micael Verdi, we think it is important for the add-on to not remove itself so that we can continue to get retention data after the prompts are complete.

Schalk, can you please make the change and add the data collection ping noted in the GA requirements doc? (I'll ping you in the location of the requirements). 

https://docs.google.com/document/d/1ndh-ZGCpDZpZ4uVxnK7WftOsxjWdD_dVrxd0G9RL06c/
Flags: needinfo?(schalk.neethling.bugs)
Latest XPI to use for all future funnelcake builds. All changes are GA related but, excludes the request in https://bugzilla.mozilla.org/show_bug.cgi?id=1348127#c110
Attachment #8860902 - Attachment is obsolete: true
Flags: needinfo?(schalk.neethling.bugs)
(In reply to pgerman from comment #110)
> After discussing with Micael Verdi, we think it is important for the add-on
> to not remove itself so that we can continue to get retention data after the
> prompts are complete.
> 
> Schalk, can you please make the change and add the data collection ping
> noted in the GA requirements doc? (I'll ping you in the location of the
> requirements). 
> 
> https://docs.google.com/document/d/1ndh-
> ZGCpDZpZ4uVxnK7WftOsxjWdD_dVrxd0G9RL06c/

If this is only for startup pings, then that is fine and can easily be accommodated for.

If this is for additional work, such as the change of calendar day ping, I am going to have to ask that you first clear this with Benjamin Sternthal, as well as Nicole Yee and Peter Dolanski. I am hesitant to include work at this late stage that will need additional testing to ensure it works as expected so close to launch day.

Currently, once the tour is complete, all about:newtab modifications are stopped and the pageMod is destroyed but, as the add-on is not uninstalled, it can keep sending pings for browser startup. We currently do not have any logic that will run if the browser is running and the calendar day changes, nor do we have logic that will run when the browser is started and the calendar day has changed(irrespective of the time elapsed since last launch). This will be additional logic that needs to be implemented.

We will then need additional time for internal testing, and QA from Softvision.

This will definitely cause the planned launch date of 27 March, 2017 to be pushed out another week.
(In reply to Schalk Neethling [:espressive] from comment #109)
> Hey all,
> 
> Can we have a preference that will only be set for the final production
> funnelcake build? i.e. not set while the funnelcakes are still being
> internally tested and QA'd

I think so.  We'd have to update the distribution.ini's before the final production build.

> If so, we need preferences set as follows:
> 
> For all funnelcakes
> 
> distribution.variation, this will be set as either contentVariationA or
> contentVariationB

What does this mean specifically?
(What do we set specifically for 110, 111, 112, and 113?)

> browser.newtab.preload, false
> 
> For the production build, we also need
> 
> onboard.environment, 'production' // this is needed for Google Analytics
> 
> Let me know, thanks!

I believe we need to update the [Preferences] section in https://github.com/mozilla-partners/funnelcake/blob/master/desktop/funnelcake111/distribution/distribution.ini (as well as 110, 112, 113).  Since we want some testing changes and a final production change, I imagine that means 2 PRs.  Do you or someone else want to create those, or tell me what specifically needs to be changed?
(In reply to Schalk Neethling [:espressive] from comment #112)
> (In reply to pgerman from comment #110)
> > After discussing with Micael Verdi, we think it is important for the add-on
> > to not remove itself so that we can continue to get retention data after the
> > prompts are complete.
> > 
> > Schalk, can you please make the change and add the data collection ping
> > noted in the GA requirements doc? (I'll ping you in the location of the
> > requirements). 
> > 
> > https://docs.google.com/document/d/1ndh-
> > ZGCpDZpZ4uVxnK7WftOsxjWdD_dVrxd0G9RL06c/
> 
> If this is only for startup pings, then that is fine and can easily be
> accommodated for.
> 
> If this is for additional work, such as the change of calendar day ping, I
> am going to have to ask that you first clear this with Benjamin Sternthal,
> as well as Nicole Yee and Peter Dolanski. I am hesitant to include work at
> this late stage that will need additional testing to ensure it works as
> expected so close to launch day.
> 
> Currently, once the tour is complete, all about:newtab modifications are
> stopped and the pageMod is destroyed but, as the add-on is not uninstalled,
> it can keep sending pings for browser startup. We currently do not have any
> logic that will run if the browser is running and the calendar day changes,
> nor do we have logic that will run when the browser is started and the
> calendar day has changed(irrespective of the time elapsed since last
> launch). This will be additional logic that needs to be implemented.
> 
> We will then need additional time for internal testing, and QA from
> Softvision.
> 
> This will definitely cause the planned launch date of 27 March, 2017 to be
> pushed out another week.

Thanks for outlining the downstream impacts of this. 

Please move forward with the start-up pings and I'll do a quick check in with some of the other stakeholders to determine the priority of the daily ping requirement.
Flags: needinfo?(schalk.neethling.bugs)
Schalk - thanks for flagging me here.

Schalk is available to help with questions but has officially been moved to other projects, per conversations with Peter D on April 3, Taipei folks would be taking over this work.
The impact of having a ping sent each day from a browser (in addition to start up):
-Allow us to see retention for users that kept the browser open over the course of several days and ensure they are counted for each day the browser is open 

The impact of having the add-on continue to stay installed after onboarding is complete:
-When the add on removes itself, we'll lose the mechanism to see retention in GA
(In reply to Aki Sasaki [:aki] from comment #113)
> (In reply to Schalk Neethling [:espressive] from comment #109)
> > Hey all,
> > 
> > Can we have a preference that will only be set for the final production
> > funnelcake build? i.e. not set while the funnelcakes are still being
> > internally tested and QA'd
> 
> I think so.  We'd have to update the distribution.ini's before the final
> production build.
> 
> > If so, we need preferences set as follows:
> > 
> > For all funnelcakes
> > 
> > distribution.variation, this will be set as either contentVariationA or
> > contentVariationB
> 
> What does this mean specifically?
> (What do we set specifically for 110, 111, 112, and 113?)

Nicole pointed me at comment 0.

> 
> > browser.newtab.preload, false

I landed this:
https://github.com/mozilla-partners/funnelcake/commit/b4074f6aee3c093fe9bc2102dadb2b79a772756e

Does that look good?
It appears that the various preferences are otherwise reflected here.
(In reply to Aki Sasaki [:aki] from comment #117)
> (In reply to Aki Sasaki [:aki] from comment #113)
> > (In reply to Schalk Neethling [:espressive] from comment #109)
> > > Hey all,
> > > 
> > > Can we have a preference that will only be set for the final production
> > > funnelcake build? i.e. not set while the funnelcakes are still being
> > > internally tested and QA'd
> > 
> > I think so.  We'd have to update the distribution.ini's before the final
> > production build.
> > 
> > > If so, we need preferences set as follows:
> > > 
> > > For all funnelcakes
> > > 
> > > distribution.variation, this will be set as either contentVariationA or
> > > contentVariationB
> > 
> > What does this mean specifically?
> > (What do we set specifically for 110, 111, 112, and 113?)
> 
> Nicole pointed me at comment 0.
> 
> > 
> > > browser.newtab.preload, false
> 
> I landed this:
> https://github.com/mozilla-partners/funnelcake/commit/
> b4074f6aee3c093fe9bc2102dadb2b79a772756e
> 
> Does that look good?
> It appears that the various preferences are otherwise reflected here.

That looks great Aki.

One thing to note from https://bugzilla.mozilla.org/show_bug.cgi?id=1348127#c109, we need the following pref set once the final funnelcakes are baked:

onboard.environment, 'production'
Flags: needinfo?(schalk.neethling.bugs)
(In reply to Schalk Neethling [:espressive] from comment #118)
> (In reply to Aki Sasaki [:aki] from comment #117)
> > I landed this:
> > https://github.com/mozilla-partners/funnelcake/commit/
> > b4074f6aee3c093fe9bc2102dadb2b79a772756e
> > 
> > Does that look good?
> > It appears that the various preferences are otherwise reflected here.
> 
> That looks great Aki.

Thanks for checking!

> One thing to note from
> https://bugzilla.mozilla.org/show_bug.cgi?id=1348127#c109, we need the
> following pref set once the final funnelcakes are baked:
> 
> onboard.environment, 'production'

Created https://github.com/mozilla-partners/funnelcake/pull/57 .
Created new builds with onboarding 1.0.6:

https://archive.mozilla.org/pub/firefox/candidates/53.0-candidates/build6/partner-repacks/funnelcake110/v3/win32/en-US/Firefox%20Setup%2053.0.exe
SHA256(repacked_builds/signed/partner-repacks/funnelcake110/v3/win32/en-US/Firefox Setup 53.0.exe)= 10b205854161ca6dfe7ba737a1a81c82df454aabdd21d94a046597619b6dfb0a

https://archive.mozilla.org/pub/firefox/candidates/53.0-candidates/build6/partner-repacks/funnelcake111/v3/win32/en-US/Firefox%20Setup%2053.0.exe
SHA256(repacked_builds/signed/partner-repacks/funnelcake111/v3/win32/en-US/Firefox Setup 53.0.exe)= 10829adc19bda88f162849eb4c16c7f3b803f95d65d7a29cf02fca6e523863c5

https://archive.mozilla.org/pub/firefox/candidates/53.0-candidates/build6/partner-repacks/funnelcake111/v3/win32/en-US/Firefox%20Setup%2053.0.exe
SHA256(repacked_builds/signed/partner-repacks/funnelcake112/v3/win32/en-US/Firefox Setup 53.0.exe)= 87b684b9b27cbf2f40592f14c237ed55653b0811177d7900f40b7e57fda38836

https://archive.mozilla.org/pub/firefox/candidates/53.0-candidates/build6/partner-repacks/funnelcake112/v3/win32/en-US/Firefox%20Setup%2053.0.exe
SHA256(repacked_builds/signed/partner-repacks/funnelcake113/v3/win32/en-US/Firefox Setup 53.0.exe)= 492a9bb1668ba35ce979ad00043e2fa5c5eb3cdc00b11023d97932f3ba2e391b

I also updated the bouncer links, which were still pointing at build5 v1.
Do these new builds include the Stub Installer or just the fix for the Sync Link?
Flags: needinfo?(aki)
I copied the stub installers forward and updated bouncer.
Flags: needinfo?(aki)
FYI there is a duplicate in the links above. You have provided 110, 111 twice, and 112. 113 is missing.
Flags: needinfo?(aki)
(In reply to Justin [:JW_SoftvisionQA] from comment #123)
> FYI there is a duplicate in the links above. You have provided 110, 111
> twice, and 112. 113 is missing.

These were manually created links. The builds should be fine.  Editing for posterity:

https://archive.mozilla.org/pub/firefox/candidates/53.0-candidates/build6/partner-repacks/funnelcake110/v3/win32/en-US/Firefox%20Setup%2053.0.exe
SHA256(repacked_builds/signed/partner-repacks/funnelcake110/v3/win32/en-US/Firefox Setup 53.0.exe)= 10b205854161ca6dfe7ba737a1a81c82df454aabdd21d94a046597619b6dfb0a

https://archive.mozilla.org/pub/firefox/candidates/53.0-candidates/build6/partner-repacks/funnelcake111/v3/win32/en-US/Firefox%20Setup%2053.0.exe
SHA256(repacked_builds/signed/partner-repacks/funnelcake111/v3/win32/en-US/Firefox Setup 53.0.exe)= 10829adc19bda88f162849eb4c16c7f3b803f95d65d7a29cf02fca6e523863c5

https://archive.mozilla.org/pub/firefox/candidates/53.0-candidates/build6/partner-repacks/funnelcake112/v3/win32/en-US/Firefox%20Setup%2053.0.exe
SHA256(repacked_builds/signed/partner-repacks/funnelcake112/v3/win32/en-US/Firefox Setup 53.0.exe)= 87b684b9b27cbf2f40592f14c237ed55653b0811177d7900f40b7e57fda38836

https://archive.mozilla.org/pub/firefox/candidates/53.0-candidates/build6/partner-repacks/funnelcake113/v3/win32/en-US/Firefox%20Setup%2053.0.exe
SHA256(repacked_builds/signed/partner-repacks/funnelcake113/v3/win32/en-US/Firefox Setup 53.0.exe)= 492a9bb1668ba35ce979ad00043e2fa5c5eb3cdc00b11023d97932f3ba2e391b
Flags: needinfo?(aki)
Email Response from Shane Tomlinson per sync URL

"I have found the problem!

OCSP stapling is a browser security feature that ensures SSL certs for a domain have not been revoked. If a browser thinks the OCSP response is invalid or out of date, it'll refuse to download the resource. The thing is, OCSP responses are only considered valid for a short period of time. Because we have to set the clock to 2 weeks in the future to test whether the FxA link works, the OCSP response is considered invalid and none of our JS/CSS is downloaded.

If I walk through funnelcake #112, incrementing the system clock by 2 days each time, I end up at May 5th before I see the Sync notification. I click on the Sync link, nothing happens.

Tellingly, if I just visit https://accounts.firefox.com in Firefox on the host machine (instead of the VM), the same problem manifests. There is nothing unique about the notification link [1], any FxA link will have the same problem as long as the date is that far in the future.

If I reset the system time to the correct current time, both [1] and plain old https://accounts.firefox.com load, on both guest and host machines.

There is no possible "fix" for this other than resetting the system clock. There is no user facing bug, as long as the user's clock is somewhat in sync with the rest of the world.

Thanks for the help Justin, your guidance made the problem really easy to find.

Shane

[1] - about:accounts?action=signup&entrypoint=uitour&utm_source=onboarding-addon&utm_medium=referral&utm_campaign=contentVariationB "
Everything it the builds from comment 124 is working as expected!
Error:

Error with firefox_onboard-1.0.6-fx.xpi
100% complete · 35.56 KB of 35.56 KB
Your add-on failed validation with 1 error.

Starting with Firefox 53, new extensions on this site can only be WebExtensions.

I can't upload the XPI via AMO, so maybe the AMO folks can help do it on their side.
Depends on: 1359267
(In reply to Justin [:JW_SoftvisionQA] from comment #125)
> Email Response from Shane Tomlinson per sync URL
> 
> "I have found the problem!
> 
> OCSP stapling is a browser security feature that ensures SSL certs for a
> domain have not been revoked. If a browser thinks the OCSP response is
> invalid or out of date, it'll refuse to download the resource. The thing is,
> OCSP responses are only considered valid for a short period of time. Because
> we have to set the clock to 2 weeks in the future to test whether the FxA
> link works, the OCSP response is considered invalid and none of our JS/CSS
> is downloaded.
> 


Some more on this. It is possible to bypass OCSP checks by modifying an about:config preference:

security.OCSP.enabled=0
Official add-on public on AMO. We are ready for production build funnelcakes! :)

https://bugzilla.mozilla.org/show_bug.cgi?id=1359267
Flags: needinfo?(nyee) → needinfo?(aki)
Production funnelcake candidates:

https://archive.mozilla.org/pub/firefox/candidates/53.0-candidates/build6/partner-repacks/funnelcake110/v4/win32/en-US/Firefox%20Setup%2053.0.exe
SHA256(repacked_builds/signed/partner-repacks/funnelcake110/v4/win32/en-US/Firefox Setup 53.0.exe)= 05222bb89ca95cdb7695879de0a33de686743fbbeb96c5b0fb447caf736f1269

https://archive.mozilla.org/pub/firefox/candidates/53.0-candidates/build6/partner-repacks/funnelcake111/v4/win32/en-US/Firefox%20Setup%2053.0.exe
SHA256(repacked_builds/signed/partner-repacks/funnelcake111/v4/win32/en-US/Firefox Setup 53.0.exe)= f92a3293206ba7db1980cb80447661a2a5dfc6321a47a498a695cff6f4086420

https://archive.mozilla.org/pub/firefox/candidates/53.0-candidates/build6/partner-repacks/funnelcake112/v4/win32/en-US/Firefox%20Setup%2053.0.exe
SHA256(repacked_builds/signed/partner-repacks/funnelcake112/v4/win32/en-US/Firefox Setup 53.0.exe)= ab646374f150d24046f224917abe3e5ee685b0fb20391d672d9cc9cf985ec59c

https://archive.mozilla.org/pub/firefox/candidates/53.0-candidates/build6/partner-repacks/funnelcake113/v4/win32/en-US/Firefox%20Setup%2053.0.exe
SHA256(repacked_builds/signed/partner-repacks/funnelcake113/v4/win32/en-US/Firefox Setup 53.0.exe)= d5bda44ccf13385225829307a9c7b96a114b2adbb16a27880f163a51d35b9ab9
Flags: needinfo?(aki)
13:30 <nicolejadeyee> hey aki. did we build the production funnelcakes with the amo xpi?
13:30 <aki> i built it with 1.0.6 in the bug. should i have downloaded it from amo?
13:31 <nicolejadeyee> yes, it should be built with the one downloaded from amo.
13:31 <nicolejadeyee> sorry i wasnt clear earlier and didnt catch this till now

Building v5.
I will start testing this now. We should complete testing by mid day tomorrow.
Justin, let's wait till we get the updated build with the amo downloaded add-on.
After we can test the production build with the stub installer in bug 1353576.
With the signed amo addon:

https://archive.mozilla.org/pub/firefox/candidates/53.0-candidates/build6/partner-repacks/funnelcake110/v5/win32/en-US/Firefox%20Setup%2053.0.exe
SHA256(repacked_builds/signed/partner-repacks/funnelcake110/v5/win32/en-US/Firefox Setup 53.0.exe)= 88bc317ae0107daee81f0f2bcb7b2ceb194954bc06cd5c38fd2735b277df2940

https://archive.mozilla.org/pub/firefox/candidates/53.0-candidates/build6/partner-repacks/funnelcake111/v5/win32/en-US/Firefox%20Setup%2053.0.exe
SHA256(repacked_builds/signed/partner-repacks/funnelcake111/v5/win32/en-US/Firefox Setup 53.0.exe)= bd26eefba5b4a059b40b944662cc9a936af2eae1d4027a0c52fe5a23924bfda4

https://archive.mozilla.org/pub/firefox/candidates/53.0-candidates/build6/partner-repacks/funnelcake112/v5/win32/en-US/Firefox%20Setup%2053.0.exe
SHA256(repacked_builds/signed/partner-repacks/funnelcake112/v5/win32/en-US/Firefox Setup 53.0.exe)= 586bdd1e9bdfb96b346e939b05c5f5a973112fb106e1d1aa39c17b7f164c034a

https://archive.mozilla.org/pub/firefox/candidates/53.0-candidates/build6/partner-repacks/funnelcake113/v5/win32/en-US/Firefox%20Setup%2053.0.exe
SHA256(repacked_builds/signed/partner-repacks/funnelcake113/v5/win32/en-US/Firefox Setup 53.0.exe)= d347d89c12094bca8d82aa943af1c3ee2b5a24d0fa8a9ec9e9e9ed2058cb8193
I am running into the issue shown in https://bugzilla.mozilla.org/attachment.cgi?id=8859358 again.
Flags: needinfo?(nyee)
Flags: needinfo?(aki)
Jorge, any idea why we are still running into this error if the builds are using the AMO downloaded XPIs?
Flags: needinfo?(nyee) → needinfo?(jorge)
I can confirm the file on AMO has been correctly signed. Can you verify that you're using the AMO file? Is the file being unpacked or modified in any way?
Flags: needinfo?(jorge)
(In reply to Jorge Villalobos [:jorgev] from comment #138)
> I can confirm the file on AMO has been correctly signed. Can you verify that
> you're using the AMO file? Is the file being unpacked or modified in any way?

The file is being unpacked, per the instructions I was given.
https://gist.github.com/escapewindow/5dba93a02bb346f5ac3677477106a46f#x-generate-full-installer-configs
Flags: needinfo?(aki)
Okay. Unpacking in itself shouldn't cause problems, but it can lead to signature failures if anything changes in the add-on directory. Not sure if this is what's happening here. First I'd like confirmation that this is the AMO file and it has a META-INF directory in it.
The unpacked addon is here: https://github.com/mozilla-partners/funnelcake/tree/master/desktop/funnelcake111/distribution/extensions/%40onboard-v2 .  It has the META-INF directory in it, and I grabbed it from https://addons.mozilla.org/en-US/firefox/addon/firefox-onboarding-tour/ .

Per the docs I then
    cd <funnelcake_repo>/desktop/
    rsync -a --delete funnelcake111/distribution/extensions/ funnelcake112/distribution/extensions/
    rsync -a --delete funnelcake111/distribution/extensions/ funnelcake113/distribution/extensions/

https://gist.github.com/escapewindow/5dba93a02bb346f5ac3677477106a46f#revising-the-builds
That all looks right, so I don't know why it's failing.

Andrew, what can be done to debug this issue further? See comment #136 and responses.
Flags: needinfo?(aswan)
Is there anything in the Browser Console?
If not pleaes needinfo me again and I'll try to reproduce it and look into it.
Flags: needinfo?(aswan)
Browser Console:


1493238232315	addons.xpi	WARN	Add-on @onboard-v2 is not correctly signed.
1493238232345	addons.xpi	WARN	Add-on @onboard-v2 is not correctly signed.
Content Security Policy: Directive ‘frame-src’ has been deprecated. Please use directive ‘child-src’ instead. (unknown)
Login data scheme type not supported: 3 ChromeProfileMigrator.js:484
Key event not available on some keyboard layouts: key=“r” modifiers=“accel,alt” id=“toggleReaderMode”  browser.xul
Key event not available on some keyboard layouts: key=“i” modifiers=“accel,alt,shift” id=“key_browserToolbox”  browser.xul
Use of getPreventDefault() is deprecated.  Use defaultPrevented instead.  246059135.js:93:44
Content Security Policy: Couldn’t process unknown directive ‘worker-src’  (unknown)
SHIELD active  selfrepair.8d056d434484b1d6ae51.js:9:8032
Flags: needinfo?(aswan)
Looks like we had an issue with unpacking the AMO copy of the addon into the git repo (the __MACOSX and amo/data directories are missing but in the signing manifest). I'll create a PR to fix that and we'll generate more funnelcake builds.
(In reply to Nick Thomas [:nthomas] from comment #145)
> Looks like we had an issue with unpacking the AMO copy of the addon into the
> git repo (the __MACOSX and amo/data directories are missing but in the
> signing manifest). I'll create a PR to fix that and we'll generate more
> funnelcake builds.

This is indeed the problem.  Though I don't see any "amo" directory, the missing ones are __MACOSX and data/media
Flags: needinfo?(aswan)
Yes I agree, the difference is just a failure to translate from my temporary unpacked dir.
Thanks Nick! Can you submit the updated XPI a 1.0.7 to AMO for Jorge to sign? Then we can respin the builds with the new AMO signed add-on and try testing again.
Flags: needinfo?(nthomas)
The fix was in the repo for the funnelcake config (see comment #146) rather than the onboarding addon, so AFAIK we don't need a v1.0.7 or AMO submission. Aki is working on spinning a set of builds with the fix.
Flags: needinfo?(nthomas)
Nick fixed the addon.
Here's a new set of builds:

https://archive.mozilla.org/pub/firefox/candidates/53.0-candidates/build6/partner-repacks/funnelcake110/v7/win32/en-US/Firefox%20Setup%2053.0.exe
SHA256(repacked_builds/signed/partner-repacks/funnelcake110/v7/win32/en-US/Firefox Setup 53.0.exe)= d8b2088f237336c228222bb6c8bc87a04449e205e0f2bc1995776539f54598b2

https://archive.mozilla.org/pub/firefox/candidates/53.0-candidates/build6/partner-repacks/funnelcake111/v7/win32/en-US/Firefox%20Setup%2053.0.exe
SHA256(repacked_builds/signed/partner-repacks/funnelcake111/v7/win32/en-US/Firefox Setup 53.0.exe)= 4335c9fd0a679a1164b9135365b5fe53ced772de7729aec844ea99805f5051a9

https://archive.mozilla.org/pub/firefox/candidates/53.0-candidates/build6/partner-repacks/funnelcake112/v7/win32/en-US/Firefox%20Setup%2053.0.exe
SHA256(repacked_builds/signed/partner-repacks/funnelcake112/v7/win32/en-US/Firefox Setup 53.0.exe)= d7f9e86f45f61038e10111658bcfb2a922220331249993c154554f46fccf823f

https://archive.mozilla.org/pub/firefox/candidates/53.0-candidates/build6/partner-repacks/funnelcake113/v7/win32/en-US/Firefox%20Setup%2053.0.exe
SHA256(repacked_builds/signed/partner-repacks/funnelcake113/v7/win32/en-US/Firefox Setup 53.0.exe)= 4ced91ee51d3f307ef0c2c78980969a0d0308f30fde23c55c45f7c913732c41a
I can verify the addon is installing as expected. I will put in some extra time tonight and finish testing this.
Everything looks good on Win 7, 8.1, and 10. I did notice however that the stub installer still is not included. I will stick around for another hour to answer questions and/or test.
Flags: needinfo?(nyee)
(In reply to Justin [:JW_SoftvisionQA] from comment #153)
> Everything looks good on Win 7, 8.1, and 10. I did notice however that the
> stub installer still is not included. I will stick around for another hour
> to answer questions and/or test.

What do you mean?
The stub installer is in the same directory, e.g. https://archive.mozilla.org/pub/firefox/candidates/53.0-candidates/build6/partner-repacks/funnelcake113/v7/win32/en-US/Firefox%20Setup%20Stub%2053.0.exe
Ok. I was under the impression that the links provided in comment 151 were suppose to include the stub installer.
Quick question Aki. Is there a difference from the 113 build in comment 151 and the one you provided in comment 154 other than the Stub Installer? I am asking to see if I should rerun these tests with the builds containing the stub installer.
Flags: needinfo?(aki)
I'm not sure what "containing the stub installer" means.  Aiui, the stub installer points at the full installer.  So as far as I know, if the stub installer installs the same build, I think we're good.
Flags: needinfo?(aki)
Ok. The "New Stub Installer" has all the fixes from https://bugzilla.mozilla.org/show_bug.cgi?id=1328445 . These changes were in the build you provided in comment 154. The default stub installer was in the builds provided in comment 151. But like you said, it should all be good. Thank you for your help and sorry for the confusion.
(In reply to Justin [:JW_SoftvisionQA] from comment #158)
> Ok. The "New Stub Installer" has all the fixes from
> https://bugzilla.mozilla.org/show_bug.cgi?id=1328445 . These changes were in
> the build you provided in comment 154. The default stub installer was in the
> builds provided in comment 151. But like you said, it should all be good.
> Thank you for your help and sorry for the confusion.

I think the links in comment 151 were the full installers, but otherwise I think this is correct.
Thanks!
Testing has been completed!!
Flags: needinfo?(agashlin)
Sounds good, we should just need to rename the stubs "Firefox Installer.exe".

I'm not sure what the ni? is for?
Flags: needinfo?(agashlin) → needinfo?(jwilliams)
Sorry Nicole informed me to comment so that you could resolved this bug.
Flags: needinfo?(jwilliams)
(In reply to Adam Gashlin [:agashlin] from comment #161)
> Sounds good, we should just need to rename the stubs "Firefox Installer.exe".
> 
> I'm not sure what the ni? is for?

Where would we make that change. i could help identify a person.
Flags: needinfo?(nyee) → needinfo?(agashlin)
(In reply to Nicole Yee (:nicoleyee) from comment #163)
> (In reply to Adam Gashlin [:agashlin] from comment #161)
> > Sounds good, we should just need to rename the stubs "Firefox Installer.exe".
> 
> Where would we make that change. i could help identify a person.

I think whoever is handling relman? Aki?
Flags: needinfo?(agashlin) → needinfo?(aki)
They're currently named "Firefox Setup Stub 53.0.exe", and have been throughout testing.  Aiui this is the standard naming convention.  Is there a reason to rename?

If we need to, we can change the simple 'sync' command [1] to multi-line rename-and-move commands per file, and/or create a v8/ set of directories with the proper naming, then change the bouncer entries to match the new names: these are each manual, and can be typo'ed.  Other people dealing with these files, who expect to see 'stub' in the stub installer filename, may have to dig to understand that "Firefox Installer.exe" is the stub installer.  These aren't hard blockers; I just want to verify that there's a solid reason here.

[1] https://gist.github.com/escapewindow/5dba93a02bb346f5ac3677477106a46f/#--ready-to-go-live
Flags: needinfo?(aki)
Given aki's feedback and that we have already completed testing, I would prefer if we kept the name as is. Unless there is an solid reasoning.
Yes please rename the stub installer. This will keep it consistent with the previous test. It's all part of making things consistent and in this case human readable.
The bouncer entries are pointing at the latest v7/ candidates.

I believe the stub installers point at the bouncer entries, which point at the full installers.
Flags: needinfo?(aki)
(In reply to Aki Sasaki [:aki] from comment #170)
> The bouncer entries are pointing at the latest v7/ candidates.
> 
> I believe the stub installers point at the bouncer entries, which point at
> the full installers.

Are the renaming of the installers mentioned above done too?

The mozilla.org team is ready to enable the funnelcake distribution and I want to make sure everything is good to go.
I was waiting for a final answer about the rename.  We have 2 yes and 2 no answers.  Who has final say?

(In reply to Verdi [:verdi] from comment #167)
> Yes please rename the stub installer. This will keep it consistent with the
> previous test. It's all part of making things consistent and in this case
> human readable.

I think this will make the naming for this funnelcake different from all other Firefox funnelcakes and releases, so it depends on what we mean by "consistent" here.
Peter has final say on this who is on PTO. Pinging Jeff.
(In reply to Aki Sasaki [:aki] from comment #172)
> I think this will make the naming for this funnelcake different from all
> other Firefox funnelcakes and releases, so it depends on what we mean by
> "consistent" here.

By consistent I mean consistent with our team's last test were we named the installer files "Firefox Installer." This matches what they do, doesn't include weird software jargon and matches the name displayed in the installer in the next step (i.e. "Firefox Installer" instead of "Firefox Setup"). Nicole and I discussed this over slack and while I feel like this is an important detail, in this case, shipping today is more important and I'm okay with skipping this if that makes shipping today possible.
I would vote for making the file name consistent with the past test, but I don't want to hold up. I am on PTO tomorrow and I wanted to ship today.
Ok.  I've renamed in the v8/ directory: https://archive.mozilla.org/pub/firefox/candidates/53.0-candidates/build6/partner-repacks/funnelcake110/v8/win32/en-US/

I can copy either the v7 (not renamed) or v8 (renamed) directories live.  Does v8 look ok?
(In reply to Aki Sasaki [:aki] from comment #176)
> Ok.  I've renamed in the v8/ directory:
> https://archive.mozilla.org/pub/firefox/candidates/53.0-candidates/build6/
> partner-repacks/funnelcake110/v8/win32/en-US/
> 
> I can copy either the v7 (not renamed) or v8 (renamed) directories live. 
> Does v8 look ok?

The names in v8 look right but when I download "Firefox Installer.exe" I get the regular stub installer and not the one were supposed to use for the test.
(In reply to Verdi [:verdi] from comment #177)
> (In reply to Aki Sasaki [:aki] from comment #176)
> > Ok.  I've renamed in the v8/ directory:
> > https://archive.mozilla.org/pub/firefox/candidates/53.0-candidates/build6/
> > partner-repacks/funnelcake110/v8/win32/en-US/
> > 
> > I can copy either the v7 (not renamed) or v8 (renamed) directories live. 
> > Does v8 look ok?
> 
> The names in v8 look right but when I download "Firefox Installer.exe" I get
> the regular stub installer and not the one were supposed to use for the test.

Oh this is because 110 is the control. Actually that one should NOT have it's name changed. Just the ones for 111, 112 and 113.
(In reply to Verdi [:verdi] from comment #178)
> (In reply to Verdi [:verdi] from comment #177)
> > (In reply to Aki Sasaki [:aki] from comment #176)
> > > Ok.  I've renamed in the v8/ directory:
> > > https://archive.mozilla.org/pub/firefox/candidates/53.0-candidates/build6/
> > > partner-repacks/funnelcake110/v8/win32/en-US/
> > > 
> > > I can copy either the v7 (not renamed) or v8 (renamed) directories live. 
> > > Does v8 look ok?
> > 
> > The names in v8 look right but when I download "Firefox Installer.exe" I get
> > the regular stub installer and not the one were supposed to use for the test.
> 
> Oh this is because 110 is the control. Actually that one should NOT have
> it's name changed. Just the ones for 111, 112 and 113.

v9 incoming...
(In reply to Aki Sasaki [:aki] from comment #180)
> Ok,
> https://archive.mozilla.org/pub/firefox/candidates/53.0-candidates/build6/
> partner-repacks/funnelcake110/v9/win32/en-US/ has the old stub name, and
> https://archive.mozilla.org/pub/firefox/candidates/53.0-candidates/build6/
> partner-repacks/funnelcake111/v9/win32/en-US/ (and 112 113) have the renamed
> stub name.
> 
> If that looks good I'll push live.

Forgot the bouncer fixes; that's done now.
Flags: needinfo?(nthomas)
That looks right to me! Thank you so much.
Ok, :aki and :verdi: we 100% ready to go live?
(In reply to Aki Sasaki [:aki] from comment #184)
> http://archive.mozilla.org/pub/firefox/releases/53.0-funnelcake110/
> http://archive.mozilla.org/pub/firefox/releases/53.0-funnelcake111/
> http://archive.mozilla.org/pub/firefox/releases/53.0-funnelcake112/
> http://archive.mozilla.org/pub/firefox/releases/53.0-funnelcake113/
> 
> I pointed bouncer at the new locations for stub and full installers.
> 
> Per
> https://gist.github.com/escapewindow/5dba93a02bb346f5ac3677477106a46f/#--
> ready-to-go-live , next steps are
> 
> * someone checks the end-to-end again
Justin, can you do one for end-to-end test again for us?


> * optimizely enabled at 1% for a few days, then will be increased to 10% to
> ship for real

Once justin signs off, we are good to go cmore.
Flags: needinfo?(jwilliams)
To be clear, the end-to-end test starts at https://www.mozilla.org/en-US/firefox/new/?f=110 etc ?
Getting right on that!
Flags: needinfo?(jwilliams)
Depends on: 1360400
(In reply to Chris More [:cmore] from comment #188)
> https://www.mozilla.org/en-US/firefox/new/?f=113

I get a 404 error when trying to download this build.
Flags: needinfo?(chrismore.bugzilla)
Status: v10 is up, without production env, and passed tests.  Bouncer is still pointing at v10.
I also built v11, with production env.  We can push this tomorrow morning.
1:13 am CST - I just tested the full flow starting from the urls in comment 188. Here's what I found:

The first time the default browser modal is displayed over top of the firstrun page or the new tab page, these pages will not load until you take action on the modal dialog. I was able to reproduce this in release 53 - filed Bug 1360445

FC110 Has the wrong download page. It uses the nighttime download page when it should use the normal download page. I commented here https://bugzilla.mozilla.org/show_bug.cgi?id=1348129#c39

FC111, FC112 and FC113 all work as expected* except that they don't have the correct firstrun page but this is expected until Bug 1360397 is fixed.  

* Add-on issues that I bring up with Schalk - notifications don't appear until the 2nd user-created new tab (3rd new tab overall) and there is an ugly flash of shifted tiles every time you create a new tab.

To summarize: 
If we can fix Bug 1360397 and address https://bugzilla.mozilla.org/show_bug.cgi?id=1348129#c39 I think we'll be ready for the final build and QA.
(In reply to Verdi [:verdi] from comment #191)
> 1:13 am CST - I just tested the full flow starting from the urls in comment
> 188. Here's what I found:
> 
> * Add-on issues that I bring up with Schalk - notifications don't appear
> until the 2nd user-created new tab (3rd new tab overall) and there is an
> ugly flash of shifted tiles every time you create a new tab.

This is all related to how Firefox caches and reports the URL. I tried absolutely everything and, when the browser opens, the url that the add-on received is normally about:blank, and not about:newtab. I was thinking of targeting about:blank as well during startup but, that made no difference for me locally and, may result in legitimate about:blank pages being modified. This is a limitation that I had to work with(and will be resolved when this tour is built into the product), and in fact, without having the preference browser.newtab.preload = false, about:newtab modifications would never happen.

The reason for the flash, can be one of two things

1. Because we have browser.newtab.preload = false, the content is not preloaded and so needs to be loaded at the time the tab is opened(this is something that will be resolved when this tour is built into the product)
2. A browser repaint as the add-on modifications are injected into the page.
Depends on: 1360462
Depends on: 1360445
(In reply to Justin [:JW_SoftvisionQA] from comment #189)
> (In reply to Chris More [:cmore] from comment #188)
> > https://www.mozilla.org/en-US/firefox/new/?f=113
> 
> I get a 404 error when trying to download this build.

Can you check again? I think that was before the new builds were up.
Flags: needinfo?(chrismore.bugzilla)
Flags: needinfo?(pdolanjski)
We're live!

Launched with 4% sample on May 2nd 2017 @ 12:55pm PST.
Status: NEW → RESOLVED
Closed: 7 years ago
Resolution: --- → FIXED
We have new builds for 53.0.2 that we can push live. Thought I should note it here, in case people want to test and otherwise make sure the new push goes out safely, rather than just leave messages in #onboardingwarroom.
I am capable of pushing live without a safety net, and it may go smoothly.  We can also do some 53.0.2 spot checks and ramp down Traffic Cop to 0% before pushing, and then ramp back up.
(In reply to Aki Sasaki [:aki] from comment #195)
> We have new builds for 53.0.2 that we can push live. Thought I should note
> it here, in case people want to test and otherwise make sure the new push
> goes out safely, rather than just leave messages in #onboardingwarroom.
> I am capable of pushing live without a safety net, and it may go smoothly. 
> We can also do some 53.0.2 spot checks and ramp down Traffic Cop to 0%
> before pushing, and then ramp back up.

Pushed.  We look ok from my end.
Based on slack chatter, distribution of this funnelcake was stopped a few hours ago.
Configs retired, no more funnelcakes will be auto-built for this.
Full cohort dates:

Start week: April 30th
End week: May 14th

Experiment ended on May 20th.

Results in the coming weeks.
Component: Custom Release Requests → Release Requests
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Creator:
Created:
Updated:
Size: