115337 - Old version of flash 4.x is used even when 6.x has the latest version installed

Status: VERIFIED FIXED

(Core Graveyard :: Plug-ins, defect, P1)

Description

[shrirang khanzode]

Ok..seen on 1212 trunk win NT.

steps:
1 Close existing browsers completely
2 Copy the attached files to your 4.x plugins folder and restart 4.x
3 Now, with 6.x recommended install/custom, you should get the flash 5.0r41 
  version installed automatically.
4 Just, launch 6.x build and load the page above
5 Do a right click and see that "About Flash Player 4..." is listed at the 
  bottom ..which is wrong.

ALSO, after step 5, if you do an "about :plugins" in the 6.x window, you will 
see that flash 5.x is listed. Now revisit the page above and right click...
see that the correct plugin (5.x) is listed. Plugin refresh works.

Comment 1

[shrirang khanzode]

Attachment: 4.0 flash for 4.x plugins folder

Comment 2

[R.K.Aa.]

something like this happens in bug 114026

Comment 3

[shrirang khanzode]

*** Bug 114026 has been marked as a duplicate of this bug. ***

Comment 4

[Peter Lubczynski]

Attachment: possible fix?

Here is a possible fix, but it's not the only way to fix this. By flipping the
internal linked list of plugins in reverse order after scanning is complete,
you'll notice that about:plugins now lists plugins in scan order (as apposed to
reverse like it is now). This means that whatever mimetype is seen FIRST (in
about:plugins) wins in a conflict situation. So, Flash 5 in the application
folder will win against Flash 4 in the 4.x folder as long as the application
folder is scanned first (which it is). I think this is the desired effect.

Comment 5

[Henry Wertz]

     I have this same problem; the Mozilla 0.96 does not have this problem, but
current builds do.  I build from source, but I have downloaded one or two recent
builds (last week or so) and they acted the same way.  
     But, if you look at the flash 5 plugin, the text"About Macromedia Flash
Player 5..." and "About Flash Player 4..." *both* are in the binary.  So, I
don't know what triggers the "Movie not Loaded"/"About Flash Player 4..."
behavior but it isn't due to an older version of flash being loaded.  (I'd
already looked for old plugins laying around, and don't believe I have any.)

Comment 6

[av (gone)]

Comment on attachment 61829 [details] [diff] [review]
possible fix?

r=av. This will also bring more pleasant user experience. Good to have untill
we implement more poweful plugin management tool.

Comment 7

[Arun Ranganathan]

OK, I'm slapping all the nomination keywords on this since I think this is an
important fix for users of the embedding branch(es) to benefit from.

Comment 8

[Peter Lubczynski]

Patrick, what do you think of this patch?

Comment 9

[av (gone)]

FYI, I applied the patch and so far don't see any abnormalities.

Comment 10

[Patrick C. Beard]

When we scan multiple directories, which directories do we scan first? 
Reversing this list will change that ordering as well. Perhaps we should 
instead change the code which builds the list to always append to the list 
rather than "pushing" on to the head of the list. You just need a head and 
tail pointer for that. If that's too much work, then, sure, I'll go along with 
this. Your choice.

Comment 11

[Patrick C. Beard]

Comment on attachment 61829 [details] [diff] [review]
possible fix?

sr=beard

Comment 12

[Peter Lubczynski]

Here's the order we scan:

1) XPCOM (through registry)
2) application plugins directory list from the directory service (bin, embedding)
3) os locations (OSX Internet-Plugins)
4) 4.x (selective)
5) JRE (optional)

The code for making the plugin/mime-type match just traverses the linked list so
since we are adding to the front of the list during scanning, we may get the
wrong plugin during usage.

I thought about adding a tail pointer but there is more than one place we add
and remove nodes and this approach of flipping the stack seemed simpler.

Comment 13

[Judson Valeski]

we do want this on the branch, but, let's bake it on the trunk for a couple of days.

Comment 14

[Peter Lubczynski]

in trunk

Comment 15

[Asa Dotzler [:asa]]

a=asa (on behalf of drivers) for checkin to 0.9.7

Comment 16

[Peter Lubczynski]

fixed0.9.7

Comment 17

[Michael Dunn]

Will plus for 094 once verified on trunk/097.

Comment 18

[shrirang khanzode]

verif on trunk 1221 win/mac/linux. I don't see any problem with this testcase 
and other plugins as well. 

Comment 19

[Michael Dunn]

Plus ing for 094 since fix verified on trunk

Comment 20

[Peter Lubczynski]

fixed0.9.4

Comment 21

[shrirang khanzode]

fixed on 0101 branch 0.9.4 as well. VERIF.

Comment 22

[shrirang khanzode]

had already verifid these bugs on 0.9.4 branch..adding keyword 'verified0.9.4'.