102042 - Separate the useragent used by http from the actual one seen by plug-ins

Status: RESOLVED WONTFIX

(Core Graveyard :: Plug-ins, enhancement, P5)

Description

[CJ Kucera]

It is possible to set the Useragent string by setting
"global.useragent.override" in user.js or prefs.js.  However, some plugins
(notably Java 1.3) query the Useragent to determine if they're loadable or not.
 Plugins should see the actual Useragent of Mozilla, while websites should see
the useragent overridden in the .js files.  (I realize this isn't strictly
speaking a bug, which is why I've filed it under "enhancements.")

Thinking about it, though, I suppose it's possible that someone might want to
set the useragent on a per-plugin basis, too.  I don't think we'll go into that
here.

I don't actually have any coding experience with Mozilla, but I'd be willing to
dive into this one to see if there's something I can do (unless there's someone
with experience in this area who'd feel like looking at it and be able to whip
up something much more quickly).

Comment 1

[Brian Nesse (gone)]

I have absolutely no idea who is responsible for useragent stuff...XP Apps
maybe? Hopefully asa knows... ->asa

Comment 2

[sairuh (rarely reading bugmail)]

plug-ins, perhaps...?

Comment 3

[Jacek Piskozub]

The most urgent need to this is to solve a crash which happens when the user
agent is changed to something like IE and Sun Java plug-in refuses to work (Sun
will not fix this - see bug 83376). Presently to spoof IE (for QA work) I have
to uninstall the Java Plugin first. 

Of courese finding the offending code in Mozilla (the actual crash is Mozilla
fault) should be simpler, but this solution is much more general than a fix for
one Java plug-in.

Comment 4

[Asa Dotzler [:asa]]

->XPApps

Comment 5

[sairuh (rarely reading bugmail)]

->plugins

Comment 6

[Andrew Hagen]

"Real one" is now actually a plugin (!), so summary --> "... actual one...."

Maybe it's impossible to fix this bug. Does Mozilla know it will be opening a
plugin ahead of time, so it could pass the actual useragent string?

Comment 7

[rubydoo123]

future for now

Comment 8

[rubydoo123]

The plug-ins triage team (av, beppe, peterl, serge and shrir) have reviewed this
issue and have made the following determination:

This is a low priority request and we will try to get to this at a later date.
Assigning to peterl, he has another bug that needs to be resolved and this fits
with that issue

Comment 9

[Brant Gurganus]

[RFE] is deprecated in favor of severity: enhancement.  They have the same meaning.

Comment 10

[Tony Mechelynck [:tonymec]]

I agree that this enhancement is desirable but low-priority.

Is it possible for, let's say, a Java applet to send back data (e.g., the UA string as it sees it) over the Web? If it is, wouldn't that defeat the UA spoofing privacy if this bug gets fixed?

Comment 11

[Josh Aas]

I don't want to do this. If you change your UA string then plugin compatibility is one consequence you might have to deal with.