3291 - [PP] mozilla.org image doesn't show up

Status: VERIFIED DUPLICATE of bug 7417

(Core :: Layout, defect, P3)

Description

[Akkana Peck]

The "star with mozilla" image on mozilla.org stopped showing up a few days ago
in Linux (either viewer or apprunner).  Still displays fine on Linux.

Comment 1

[troy]

Sounds like a problem with transparent images

Comment 2

[rubydoo123]

could this be related to 3291?

Comment 3

[michaelp]

is this still a problem?

Comment 4

[Christine Hoffman]

Reassigning qa contact to [e-mail] (image rendering).

Comment 5

[Eli Goldberg]

Resolving as fixed using 3.15.99 Linux build.

Specifically, verified that the image appears properly on 3.15.99 Linux build,
and does not appear using the 2.21.99 build (closest to 2.25.99 build that
readily available.)

Akkana hasn't seen it recent, either, and notes that it's been a sporadic bug.
(She adds that she'll re-open it if she sees it again.)

Comment 6

[Akkana Peck]

This has reappeared in the 3/17 dogfood builds.  Sar saw it on optimized builds,
I saw it on my debug build.

Comment 7

[pnunn]

Here's some more info on the bug:

If you save a copy of the images on mozilla.org to a local
file (using a 4.5 browser) and view them using viewer or
apprunner, they display.

If you go to the directory where the images are stored on
moz.org, http://www.mozilla.org/images/, they will
not display. If you place a break point on IL_GetImage() in
mozilla/modules/libimg/src/if.cpp, the image is never requested
from the image library.

It is sounding like a netlib bug.

Comment 8

[Gagan]

Netlib is not getting the request for that image (images/mozbanner.gif) Who
should be sending that? Assigning to Michaelp for now...

Comment 9

[michaelp]

troy, does layout make image requests (the info from pam should be useful)?

Comment 10

[michaelp]

troy, since this is a linux thing, ramiro can help (in how own words) as a
"proxy debugger."

Comment 11

[leger]

This bug is on M3 milestone which has passed.  Putting on Blank milestone to be
reassigned to a current or future milestone.

Comment 12

[kipp]

Attachment: simple test that just loads the top banner image; on linux this fails because netlib tries to write "GIF" to the server after sending the GET request

Comment 13

[Suresh Duddi (gone)]

It worked for me with todays build on linux. Anything specific I am missing.

Comment 14

[Suresh Duddi (gone)]

I am going to mark it WORKSFORME. Please reopen if you see again.

Comment 15

[Eli Goldberg]

Verifying as WORKSFORME until akkana sees this again, so the cycle can start
again. ;)

Comment 16

[Akkana Peck]

It's baaack again, in the 5/17 build.

Comment 17

[Suresh Duddi (gone)]

It worked for me with Friday 5/20 night tree.

Comment 18

[Akkana Peck]

It's back again!

Note to anyone thinking WORKSFORME: trying it once or twice doesn't really test
this bug.  It's clearly some sort of timing problem -- it comes and goes, and
doesn't happen on all machines, and seems dependant on variables such as the
speed of the machine and the speed of the net (a fast machine and a slow network
seem to make it more likely to show up).  Hopefully necko will cure it; but
until then, I think we'd better keep it open unless someone checks in something
that they actively think will fix the problem.

Comment 19

[Simon Fraser [no longer active]]

This must be a timing bug. It crops up every otherr day or so. Clearing
resolution.

Comment 20

[pnunn]

I've got an idea. reassigning to self.
-pn

Comment 21

[chris hofmann]

i'm going to move this to m7.  lets get this resolved
early in m7 development, and if it comes early enough
track me down and we can get it on the m6 branch.

Comment 22

[pnunn]

This *might* be related to #6529.
-pn

Comment 23

[Steve Lamm]

This might also be related to bug #7417.

Comment 24

[pnunn]

This is sounding more and more like 7417.
I'm marking it as a duplicate.
-pnunn

*** This bug has been marked as a duplicate of 7417 ***

Comment 25

[Eli Goldberg]

(Rubber-stamping as verified, due to high likelihood of prompt discovery oshould
the duplication resolution be erroneous.)