288907 - Flash plugin doesn't set http-accept header

Status: RESOLVED WONTFIX

(Core Graveyard :: Plug-ins, defect)

Description

[Sam Minnee]

User-Agent:       Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.7.6) Gecko/20050225 Firefox/1.0.1
Build Identifier: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.7.6) Gecko/20050225 Firefox/1.0.1

In Win/IE and Safari, installing the Flash plug-in forces the browser to include
application/x-shockwave-flash in the HTTP Accept header.  This provides an
elegant way to do server-side flash detection.

Firefox doesn't do this, presumably because there is no way for the Flash plugin
to tell Firefox to do this.

Obviously, there's going to be some confusion over whether the responsibility to
fix it lies with Mozilla or Macromedia.  Is it possible for Firefox to operate
in a similar manner to Safari and Internet Explorer in this regard?

Reproducible: Always

Steps to Reproduce:
http://dev.silverstripe.com/play/flashtest.php

* Visit the URL in IE.  application/x-shockwave-flash is there
* Visit the URL in Firefox.  application/x-shockwave-flash is not there

Comment 1

[timeless]

well, we could send all the mime types for all the plugins we accept ... we do
know what they are (after we first load them...)...

Comment 2

[Ray Schauer]

Macromedia requests that the mime-type “application/x-shockwave-flash” is added
to the HTTP-ACCEPT header when Flash Player 6 release 65 or later is installed
on all available browsers.

I'm not sure if this is a Firefox issue, or if this is Macromedia not properly
instructing Firefox to add this to the Accept header. This does limit developers
to relying on client side detection since it appears IE and Safari are the only
main stream browsers to adhere (IE loses the setting on refresh forcing the use
of a cookie to persit detection states). 

Comment 3

[Sam Minnee]

timeless@myrealbox seems to have said that it *is* a Firefox issue, inasmuch as
it's perfectly doable without Macromedia's intervention.

Sending the mime types for all installed plug-ins seems to be of practical
benefit, as well as being in keeping with the 'meaning' of the Accept: header.

Comment 4

[Boris Zbarsky [:bzbarsky]]

> Sending the mime types for all installed plug-ins seems to be of practical
> benefit

It also means sending about 3-4kb of headers with every single network request.
 The flash plugin lists 2 content types.  The Java plugin lists a dozen or so. 
Things like mplayer list about 35, etc...  This has all been discussed before
and decided against; I suggest reading the existing bugs on the Accept header.

Comment 5

[Ray Schauer]

While I see the draw backs of sending mime-types for everything installed, flash
is a tool with a very large install base. Server side detection would be great.

Comment 6

[Gervase Markham [:gerv]]

People seem to have got along happily with client-side detection of Flash for
years now. Again, read all the bugs on Accept:. The current header is a
carefully-worked compromise between sending everything and sending */*.

Gerv

Comment 7

[Jaime Mitchell (use [e-mail] for email)]

*** Bug 299089 has been marked as a duplicate of this bug. ***

Comment 8

[Gervase Markham [:gerv]]

I was under the impression that IE doesn't do this, but it turns out that it
does, on the first request only. Given that Firefox doesn't ship by default with
Flash, what should probably happen is that the Flash XPI should cause the header
to be added. This probably would mean a pref system so extensions etc. can add a
pref and all the prefs are concatenated onto the Accept: header...

I'm going to reopen this bug, but I'm not promising to fix it.

Gerv