625415 - Position is lost after a few seconds

Status: VERIFIED FIXED

(Core :: DOM: Geolocation, defect)

Description

[u279076]

Test Case:
https://litmus.mozilla.org/show_test.cgi?id=12056

Test Page:
http://hg.mozilla.org/qa/litmus-data/raw-file/tip/firefox/geolocation/position.html

The test page above is used in the test case above to check that we can share the location.  It just simply prints out the tester's coordinates once a location is shared.  However, there seems to be a regression because the location is lost after a few seconds.

Some preliminary regression testing indicates the last time this worked correctly was in Firefox 4.0b6.

Comment 1

[u279076]

This regressed between 20100831 and 20100901.

Comment 2

[u279076]

I suspect something here caused the regression:
http://hg.mozilla.org/mozilla-central/rev/0d8d6369655d

Comment 3

[Josh Matthews [:jdm]]

The reason this occurs is due to the SetTimeoutTimer() call that was added at the bottom of nsGeolocationRequest::SendLocation.  Doug, what's the rationale behind that?  I'm having trouble coming up with why we want a timer to fire after we've already sent the location.

Comment 4

[Doug Turner (:dougt)]

oic. that SetTimeooutTimer should only be called if the request was created by a WatchPosition call.

Comment 5

[Sheila Mooney]

Josh, Doug does this need to block. Can we make a call so we can get it out of the nomination queue. Who would work on it?

Comment 6

[Doug Turner (:dougt)]

Attachment: patch v.1

something like this...

Comment 7

[Josh Matthews [:jdm]]

Comment on attachment 506903 [details] [diff] [review]
patch v.1

Darn, I had this same patch sitting in my queue two weeks ago, and I thought I put it up here.  r=me as long as you get rid of the default parameter; I don't see any need for that.

Comment 8

[Doug Turner (:dougt)]

http://hg.mozilla.org/mozilla-central/rev/f169d30d28a5