1220261 - very difficult to update service worker script during local development

Status: RESOLVED WORKSFORME

(Core :: DOM: Service Workers, defect)

Description

[Ben Kelly [:bkelly, not reviewing]]

I believe the new update algorithm has landed, but its still very difficult to do local development of a service worker script.  I find I am having to go so far as to manually wipe quota manager storage, unregister the SW, and restart the browser to get the SW script to update.

Comment 1

[Ben Kelly [:bkelly, not reviewing]]

I guess the issue is that the new service worker can be stuck in the waiting state until all current windows are closed and the existing service worker stops.  I guess this is per spec, but annoying for developers.

Does this sound right Catalin?

Comment 2

[Ben Kelly [:bkelly, not reviewing]]

This time I couldn't get it to go to waiting at all.  I did this:

1) unregistered
2) wiped storage
3) refreshed page again

And I got the exact same cache name in about:serviceworkers I did before step 1.  That seems bogus to me.

Comment 3

[Ben Kelly [:bkelly, not reviewing]]

Oh... In comment 2 I had a syntax error in my script.  But with an existing service worker the .register() resolved success and I got no console reports indicating the error.  Sigh...

Comment 4

[Ben Kelly [:bkelly, not reviewing]]

Catalin writes in IRC:

 <catalinb> bkelly: so it looks like when you reload, our controlled counter doesn't reach 0, which would explain why the waiting worker doesn't replace the active worker.

Comment 5

[Cătălin Badea (:catalinb)]

(In reply to Ben Kelly [:bkelly] from comment #1)
> I guess the issue is that the new service worker can be stuck in the waiting
> state until all current windows are closed and the existing service worker
> stops.  I guess this is per spec, but annoying for developers.
> 
> Does this sound right Catalin?

Yes, I also checked this in chrome, same behaviour.

Comment 6

[(no longer active)]

Is this caused because we put the controlled document in the bf-cache?

Comment 7

[Ben Kelly [:bkelly, not reviewing]]

This has been working better for me.  I still have some unexpected behavior, but its less frequent and seems isolated to e10s.  I opened a separate (non blocking) bug for that.  See bug 1225350.

Comment 8

[Ben Kelly [:bkelly, not reviewing]]

Bill, are seeing recent problems here?  I saw your recent cc additions.  A lot has changed for the better since this bug.

Comment 9

[Bill Walker [:bwalker] [@wfwalker]]

Hi Ben, thanks for checking in -- in our workshop today we're seeing some hiccups here, but it's nothing you don't know about. I probably didn't need to CC everybody on this old bug :)