1174140 - Some service worker info properties are empty in the list of service workers shown by about:sw

Status: RESOLVED FIXED

(Core :: DOM: Service Workers, defect)

Description

[José Antonio Olivera Ortega [:jaoo]]

STR

- Open a page that registers a sw.
- Open about:sw

Expected

Service worker info properties have value.

Actual

Some of the properties are empty such as 'Current Worker URL' and 'Active Cache Name'.

Note: Everything is fine after quitting and starting the browser again. It happens also on B2G.

Feel free to move this bug to v2, or v3 please.

Comment 1

[José Antonio Olivera Ortega [:jaoo]]

This seems a regression. I've been poking around but I'm a bit lost. Could you give me any hint please? Thanks!

Comment 2

[(no longer active)]

This doesn't block the initial release of service workers.

Comment 3

[Noemí Freire (:noemi)]

It affects b2g so moving to v2. Thanks!

Comment 4

[Andrea Marchesini [:baku]]

I spoke with jaoo on IRC. Soon we will have a patch for this bug.

Comment 5

[José Antonio Olivera Ortega [:jaoo]]

While working on other about:sw issues I've noticed that this issue reproduces when about:sw loads in the parent process. about:sw should load always on the child process when e10s mode is on. What could cause about:sw page to not load in the child process when e10s mode is on?

Comment 6

[(no longer active)]

Bug 1155153 covers making about:sw work well in e10s...

Comment 7

[José Antonio Olivera Ortega [:jaoo]]

(In reply to Ehsan Akhgari (not reviewing patches, not reading bugmail, needinfo? me!) from comment #6)
> Bug 1155153 covers making about:sw work well in e10s...

I'm aware bug 1155153 covered that (it added the URI_CAN_LOAD_IN_CHILD flag for about:sw page) but the thing is that I'm printing the pid of the calling process at ServiceWorkerManager::PropagateSoftUpdate() and sometimes I see the pid of the parent process (even if e10s mode is on). When that happens the issue in this bug reproduces.

Comment 8

[(no longer active)]

Hmm, I thought URI_CAN_LOAD_IN_CHILD is enough to make the about page load in the child process...

Bill, is that not the case?

Comment 9

[Bill McCloskey [inactive unless it's an emergency] (:billm)]

You additionally need to set URI_MUST_LOAD_IN_CHILD. Otherwise it will load in the parent if the current tab is already running in the parent.

Comment 10

[José Antonio Olivera Ortega [:jaoo]]

(In reply to Bill McCloskey (:billm) from comment #9)
> You additionally need to set URI_MUST_LOAD_IN_CHILD. Otherwise it will load
> in the parent if the current tab is already running in the parent.

I just set that flag and it seems the issue is fixed. The missing properties are back now. I'll add this flag and attach a patch here. Thanks!

Comment 11

[José Antonio Olivera Ortega [:jaoo]]

Attachment: MozReview Request: Bug 1174140 - Some service worker info properties are empty in the list of service workers shown by about:sw. r=billm

Bug 1174140 - Some service worker info properties are empty in the list of service workers shown by about:sw. r=billm

Comment 12

[Bill McCloskey [inactive unless it's an emergency] (:billm)]

Comment on attachment 8630349 [details]
MozReview Request: Bug 1174140 - Some service worker info properties are empty in the list of service workers shown by about:sw. r=billm

https://reviewboard.mozilla.org/r/12707/#review11303

Ship It!

Comment 13

[Pulsebot]

https://hg.mozilla.org/integration/mozilla-inbound/rev/39dd5d40a9e3

Comment 14

[Carsten Book [:Tomcat]]

https://hg.mozilla.org/mozilla-central/rev/39dd5d40a9e3

Comment 15

[Noemí Freire (:noemi)]

Hi,

Just checked on m-c (9b902b7669ae revision) in Nightly and all the following info properties are currently properly shown in about:sw once the sw is registered:
* "Scope"
* "Script Spec"
* "Current Worker URL"
* "Active Cache Name"

once an update process is launched, "Waiting Cache Name" property is also properly shown in about:sw

a new property called "Push Endpoint" is now appearing with "Waiting..." value. We will dig into this new property and try to check its behavior once we know how it should work

I'll check it on b2g once Bug 1181351 lands. Thanks!