789466 - Report crashes with prerelease.dogfood.id if available

Status: RESOLVED FIXED

(Firefox OS Graveyard :: General, defect)

Description

[Alex Keybl [:akeybl]]

If defined, we should use prerelease.dogfood.id as the email ID for crashes. It will be the asset tag number of the dogfooding phone, and will allow us to get in touch with users affected by a specific crash.

We may want to pull this code out prior to release, to prevent privacy concerns.

Comment 1

[Alex Keybl [:akeybl]]

Not sure how we want to handle this - it's basically pre-release only, but is a requirement for dogfooding, which is a requirement for shipping.

Nomming to see what triage thinks.

Comment 2

[Jason Smith [:jsmith]]

(In reply to Alex Keybl [:akeybl] from comment #1)
> Not sure how we want to handle this - it's basically pre-release only, but
> is a requirement for dogfooding, which is a requirement for shipping.
> 
> Nomming to see what triage thinks.

Is a work-around for dogfooding possible for the email ID? That's probably the big factor for determining if it blocks or not.

Comment 3

[Andrew Overholt [:overholt]]

While we wouldn't hold the release for this (the definition of a blocker we've been using in triage), it seems important enough so let's mark it at the very least so we don't forget to remove any code we add for dogfooding.

Comment 4

[Alex Keybl [:akeybl]]

(In reply to Jason Smith [:jsmith] from comment #2)
> Is a work-around for dogfooding possible for the email ID? That's probably
> the big factor for determining if it blocks or not.

I don't believe there's any alternative to ensuring 2-way conversations for crash reports.

Comment 5

[Alex Keybl [:akeybl]]

Thanks for helping out here Hub!

Comment 6

[Lukas Blakk [:lsblakk] use ?needinfo]

We're looking at getting devices in the coming week and I'm wondering how this is coming along, this is something we'll want to test out next week if possible to confirm that a phone's config pref that's set with a prerelease.dogfood.id is being handed back to us along with crash report data. 

Also - not sure if this is the place to ask but will there be a dialog after a crash asking for user input?

Comment 7

[Jason Smith [:jsmith]]

(In reply to Lukas Blakk [:lsblakk] from comment #6)
> 
> Also - not sure if this is the place to ask but will there be a dialog after
> a crash asking for user input?

See https://github.com/mozilla-b2g/gaia/issues/4149 for the proposal for the crash UI.

Comment 8

[Hubert Figuiere [:hub]]

It is blocked by 777145. Without the former, not crash report will ever be sent.

Comment 9

[Hubert Figuiere [:hub]]

prerelease.dogfood.id is a preference value?

What annotation field shall I use for the crash reporter.

Comment 10

[Alex Keybl [:akeybl]]

(In reply to Hub Figuiere [:hub] from comment #9)
> prerelease.dogfood.id is a preference value?

We'll be using it as a preference value, although it is not a default pref. Do you foresee a problem there?

> What annotation field shall I use for the crash reporter.

Whichever field is typically used for email address can be instead used for this ID.

Comment 11

[Hubert Figuiere [:hub]]

Attachment: Bug 789466 - Report crashes with prerelease.dogfood.id if available.

Comment 12

[Hubert Figuiere [:hub]]

The change in b2g/app/b2g.js will be skipped. I forgot to remove it when submitting the patch, but it illustrated the use to reviewers.

Comment 13

[Jason Smith [:jsmith]]

Comment on attachment 665149 [details] [diff] [review]
Bug 789466 - Report crashes with prerelease.dogfood.id if available.

I probably can't conduct the review on this.

Comment 14

[Hubert Figuiere [:hub]]

(In reply to Jason Smith [:jsmith] from comment #13)
> Comment on attachment 665149 [details] [diff] [review]
> Bug 789466 - Report crashes with prerelease.dogfood.id if available.
> 
> I probably can't conduct the review on this.

Who can beside shell.js "owner" ?

Comment 15

[[:fabrice] Fabrice Desré]

Comment on attachment 665149 [details] [diff] [review]
Bug 789466 - Report crashes with prerelease.dogfood.id if available.

Review of attachment 665149 [details] [diff] [review]:
-----------------------------------------------------------------

::: b2g/chrome/content/shell.js
@@ +109,5 @@
>     },
>  
>    start: function shell_start() {
> +
> +    let dogfoodId = Services.prefs.getCharPref('prerelease.dogfood.id');

doesn't this throw if the pref is not set at all?

@@ +112,5 @@
> +
> +    let dogfoodId = Services.prefs.getCharPref('prerelease.dogfood.id');
> +    if (dogfoodId != "") {
> +      let cr = Cc["@mozilla.org/xre/app-info;1"]
> +        .getService(Ci.nsICrashReporter);

nit: align .getService with the [ on the previous line

Comment 16

[Hubert Figuiere [:hub]]

Attachment: Bug 789466 - Report crashes with prerelease.dogfood.id if available.

Addressed Fabrice comment. Remove the prefs change as well.

Comment 17

[Hubert Figuiere [:hub]]

https://hg.mozilla.org/integration/mozilla-inbound/rev/5caee20d2dc6

Comment 18

[Ed Morley [:emorley]]

https://hg.mozilla.org/mozilla-central/rev/5caee20d2dc6