133864 - Back end work to enable Plugin Finder Service Preference

Status: VERIFIED FIXED

(Core Graveyard :: Plug-ins, defect, P3)

Description

[Gregg Landskov (gone)]

This work is tied to the UI work for the Plugin Finder Service (PFS) Preference.
 Peter Lubcynski is going to do this work.

If PFS pref is on, use new system.  If PFS pref is off, use legacy system.

http://bugzilla.mozilla.org/show_bug.cgi?id=133863

Comment 1

[Gregg Landskov (gone)]

depends on 133863

Comment 2

[Peter Lubczynski]

I'm not sure the dependancy is right, might be the other way around. FE work
does not preclude this bug from being fixed.

Comment 3

[rubydoo123]

adding nsbeta1+ per ADT meeting

Comment 4

[Peter Lubczynski]

Attachment: first try at a patch to windows default plugin

Here's a patch that will always send pluginspage/pluginurl to get_plugin.cgi
unless the pref "pref.application.use_ns_plugin_finder" is set to false. I'm
using the NPAPI to get the service manager.

Comment 5

[av (gone)]

Comment on attachment 77036 [details] [diff] [review]
first try at a patch to windows default plugin

r=av

Comment 6

[Jaime Rodriguez, Jr.]

Adding ADT2 to Status Whiteboard, becasue this blocks bug 133863, which is
marked as ADT2.

Nominating for Mozilla 1.0.

Comment 7

[Jaime Rodriguez, Jr.]

What are the plans for landing this one? Is this gonna make Mozilla 1.0 RC1?

Comment 8

[Peter Lubczynski]

Attachment 77036 [details] [diff] is okay, but I think Gregg wanted some more changes. However,
he's on vacation this week.

Comment 9

[Jaime Rodriguez, Jr.]

Are gregg's changes the gating factor on resolving this one?

Comment 10

[Peter Lubczynski]

I have a patch in my tree that I'm still testing with what I think he wants.
However, I'd like to get sign-off before it's checked in.

Comment 11

[kmurray1115]

Left a vm on Peter's SD phone. Peter, call me when you have a chance and we'll
go through this.

Comment 12

[Peter Lubczynski]

Attachment: patch v.2

New patch features:

* defaults to old behavior if pref is missing
* matches pref name in bug 133863
* removes get_plugin.cgi and jvm.html entirely. We will now only do a simple
window.open() on plug-in_finder.cgi with pluginspage/pluginsurl
* sends pluginspage/pluginsurl if javascript is turned off

Comment 13

[Peter Lubczynski]

Attachment: patch v.3

New patch:
 * fixes one case when when we sent pluginspage/pluginsurl despite pref being
off
 * fixes extra 'blank window' problem

Gregg, could you update the PFS to return the right XPI's for Flash, Real, Java
for the right browser? Java currently returns a "not found" page for:
http://cgi.netscape.com/cgi-bin/plug-in_finder.cgi?mimetype=application/x-java-vm

Comment 14

[Jaime Rodriguez, Jr.]

Aruner - Is the java issue Peter mentions above, related to us having a not
having a later java up on the server for mozilla?

Comment 15

[av (gone)]

Comment on attachment 79433 [details] [diff] [review]
patch v.3

We should at least change a comment about Mozilla returning
nsIServiceManagerObsolete. QI for nsIServiceManager is not necessary but still
can be done. r=av.

Comment 16

[Peter Lubczynski]

Attachment: patch v.4

Comment 17

[av (gone)]

Comment on attachment 79519 [details] [diff] [review]
patch v.4

r=av

Comment 18

[Marc Attinasi]

Comment on attachment 79519 [details] [diff] [review]
patch v.4

sr=attinasi

Comment 19

[Jaime Rodriguez, Jr.]

Pls land this on the trunk and the branch tonight.

Comment 20

[Peter Lubczynski]

patch in trunk and branch (8am builds), marking FIXED.

Comment 21

[Randell Jesup [:jesup] (needinfo me)]

So far as I can tell, no one from drivers approved this bug for checkin to the
branch.

Comment 22

[Peter Lubczynski]

sorry! forgot Jaime doesn't speak for Drivers. Do you want me to back it out of
the branch?

Comment 23

[Jaime Rodriguez, Jr.]

Unless Mozilla really has a strong objection to this one. I'd suggest we leave
it "IN", because it is really needed by Netscape.

Comment 24

[timeless]

jaime: that's interesting but it is also the absolutely wrong thing to say.
netscape is not and should not be driving the mozilla branch.
what someone should do instead is provide reasons that mozilla.org would want 
this feature.

Comment 25

[Peter Lubczynski]

approval after the fact via e-mail:
On Wed, Apr 17, 2002 at 08:10:00AM -0700, Peter Lubczynski wrote:

> Drivers,
> 
> Sorry I checked in this patch to the 1.0 branch before getting your 
> approval. I miss took adt1.0.0+ for complete approval. Would you like me 
> to back it out?


Drivers think it's fine.  Thanks for bringing it to our attention.

Mike

---------------------
Also since the default plugin is "plug-able", I think Mozilla can make their
own. Embedding vendors are shipping their own as well.

Comment 26

[Jaime Rodriguez, Jr.]

Shrir/Greggl - Pls verify that the Plugin Finder works as designed.

Comment 27

[shrirang khanzode]

I checked out this fix. 
There are two bugs open for the new issues: bug 138500 and bug 139529. Other 
than checking the new logic of the default plugin , I cannot verify that the PFS 
is completely working fine unless these two bugs are closed and new xpi wrapped 
installers for major tier1 plugins or at least some plugins are hosted on PFS 
website. 
As far as this bug goes, the default plugin logic works nicelty now overriding 
the web author's "pluginspage/pluginsurl" attrib and takes the user to nscp 
plugin download page. adding "verified1.0.0" kwd.

Comment 28

[rubydoo123]

adding dependency

Comment 29

[shrirang khanzode]

this was long done..verif on trunk.