583010 - Some images refuse to load when page is reloaded

Status: VERIFIED WORKSFORME

(Core :: Graphics, defect)

Description

[Shane Bundy]

User-Agent:       Mozilla/5.0 (Windows; Windows NT 5.1; rv:2.0b3pre) Gecko/20100729 Minefield/4.0b3pre
Build Identifier: Mozilla/5.0 (Windows; Windows NT 5.1; rv:2.0b3pre) Gecko/20100729 Minefield/4.0b3pre

When a Facebook page is reloaded, some images do not load.

Reproducible: Sometimes

Steps to Reproduce:
1. Go to a Facebook user's page
2. Reload the page
Actual Results:  
Some images do not show.

Expected Results:  
All images should show as normal.

Reloading the page again solves the problem, but not permanently.

Comment 1

[Shane Bundy]

Attachment: A screenshot describing the problem.

Comment 2

[Shane Bundy]

Attachment: A screenshot describing what happens when the page is reloaded.

Comment 3

[Joe Drew (not getting mail)]

Shane, did this just start as of tonight's nightly?

Comment 4

[Boris Zbarsky [:bzbarsky]]

Could be a dup of the regression from the redirect changes.

Comment 6

[dolphinling]

This did start as of tonight's nightly. Also, in case it wasn't obvious, it affects nearly every page, not just facebook.

What bug is the redirect changes and what bug is the regression?

Comment 7

[Boris Zbarsky [:bzbarsky]]

Bug 513086.  There was no separate bug for the regression, just a followup fix pushed earlier today.

Comment 8

[dolphinling]

I'm seeing this on images that are not redirected at all, so I think bug 572520 (make all imgIDecoderObserver notifications asynchronous) is a more likely candidate.

Comment 9

[Boris Zbarsky [:bzbarsky]]

Though "redirected" need not be just HTTP redirects... it can include stuff with proxies, etc, etc.  We'll know once someone tests a build with the redirect fix, I guess.

Comment 10

[Honza Bambas (:mayhemer)]

(In reply to comment #8)
> I'm seeing this on images that are not redirected at all, so I think bug 572520
> (make all imgIDecoderObserver notifications asynchronous) is a more likely
> candidate.

It is not, this is clearly what is described in bug 513086 comment 79 and fixed by patch in comment 87.

(We had to open a new bug for that regression, this bug could then be easily closed as dup of that)

Comment 11

[Joe Drew (not getting mail)]

So, are people still seeing this bug since the redirect fix landed? Honza's comment 10 implies that it should be fixed now.

Comment 12

[dolphinling]

This is indeed fixed for me now.

Comment 13

[Honza Bambas (:mayhemer)]

Fixed by attachment 461244 [details] [diff] [review].