475685 - Seeking in media resources is very slow due to bug in liboggz

Status: RESOLVED DUPLICATE of bug 469408

(Core :: Audio/Video, defect, P2)

Description

[cajbir (:cajbir)]

Attachment: liboggz fix for guessing seek position

See bug 469408 comment 4. Bug 449159 introduced a bug into liboggplay where the guess for the byte position to seek too is often incorrect. This causes many unnecessary byte range requests when seeking via HTTP and slows seeks down a lot.

Comment 1

[cajbir (:cajbir)]

This should be blocking1.9.1+ I think due to the impact it has on seek times.

Comment 2

[Robert O'Callahan (:roc) (email my personal email if necessary)]

Is this patch ready for review?

Comment 3

[cajbir (:cajbir)]

Comment on attachment 359237 [details] [diff] [review]
liboggz fix for guessing seek position

Robert, are you happy to review this change or would you prefer I try and get an annodex person to look at it?

Comment 4

[Robert O'Callahan (:roc) (email my personal email if necessary)]

Comment on attachment 359237 [details] [diff] [review]
liboggz fix for guessing seek position

Ok to land. But please get an annodex person to review it upstream as well. Thanks.

Comment 5

[cajbir (:cajbir)]

Preparing this for landing tonight I noticed that all the tests pass, but is it my imagination or does the test that plays the space invaders sound have the sound cutoff? It seems to play about twice as long without this patch. I'll investigate, in the meantime I've removed the needs landing whiteboard.

Comment 6

[cajbir (:cajbir)]

Bug 477899 contains the fix that the fix to this bug triggers. After requesting the duration liboggplay seeks back to a position that is past the first part of the audio sample.

Comment 7

[Chris Pearce [:cpearce (Not reading bugmail)]]

(In reply to comment #6)
> Bug 477899 contains the fix that the fix to this bug triggers. 

Can this land now?

Comment 8

[Chris Pearce [:cpearce (Not reading bugmail)]]

Patch in bug 469408 contains the fix in doublec's patch. We can take the patch in bug 469408 instead of the one here.

Comment 9

[Robert O'Callahan (:roc) (email my personal email if necessary)]

I'm making this a dup of 469408 because that's where this will get fixed.