83229 - Referer: pref names in code and all.js do not match

Status: VERIFIED FIXED

(Core :: Networking: HTTP, defect)

Description

[Ken Weaverling]

From Bugzilla Helper:
User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:0.9) Gecko/20010505
BuildID:    2001050515

Mozilla doesn't send a referer field for graphics it requestes to loads. Other
browsers send the page that the graphic is referenced in as the referer.

This breaks my counter app, which takes the referer and determines what value to
increment. An example is at http://www.weaverling.org/atc/ under the Amtrak
sign.  If it says error, it means that the referer field on the request is NULL.
This works from all other browsers I've tried, but not mozilla.


Reproducible: Always
Steps to Reproduce:
1.Load any page with graphics
2.View web server access logs, look at referer field
3.Field is empty

Actual Results:  Referer field not sent with image requests

Expected Results:  Referer field should be sent by default with image requests
from pages

Additional info from Gervase Markham <[e-mail]> from
article [e-mail] in netscape.public.mozilla.qa.browser:

Well, doh us. We have a pref for this:

pref("network.sendRefererHeader",2); // 0=don't send any, 
1=send only on clicks, 2=send on image requests as well

But the code says:
http://lxr.mozilla.org/seamonkey/source/netwerk/protocol/http/src/nsHttpHandler.c
pp
922     if (bChangedAll || PL_strcmp(pref,
"network.http.sendRefererHeader") == 0)
923         mPrefs->GetIntPref("network.http.sendRefererHeader", (PRInt32
*) &mReferrerLevel);

However, the default is -1 ("All") so there's something a bit weird going on here.

File a bug in Networking: HTTP, include the above text, and CC me :-)

Comment 1

[Chris Hiner]

Dup of Bug 74284?  Not sending referer on any embedded objects.
See also Bug 76723 which matches this better, and is a dup of 74284.

Comment 2

[Boris Zbarsky [:bzbarsky]]

Note that bug 74284 is fixed as of May 5, while this bug is present in a May 
15th build.

Setting status to NEW.

Comment 3

[Chris Hiner]

2001050515 is a May 5th build, and it's from the 0.9 branch.  I suggest trying
it with a recent nightly and seeing if it still is a problem.

Comment 4

[Gervase Markham [:gerv]]

Look, whether it's still a problem or not, we're attempting to read a 
non-existent pref. We should fix that, and _then_ see if it's still a problem.

This is a one-line fix to all.js. If no-one else bothers, I'll do it on 
Thursday.

Gerv

Comment 5

[benc]

+regression

So this feature works now, but the pref is set wrong?

Can people fix this w/ the workaround of hacking prefs.js, so they can verify
that bug 74284 is fixed?

Comment 6

[Ken Weaverling]

I just downloaded a fresh copy of build 2001053004 (er, from today) and
installed it on a fresh, never-been-touched-by-mozilla install, tried it, and
the default setup works fine. Referer is sent with image requests.

Mozilla/5.0 (Windows; U; Win 9x 4.90; en-US; rv:0.9+) Gecko/20010530

I can't speak for whether or not the prefs to turn that behavior off or not
works now, but the default behavior is "sane" at least.

Comment 7

[Gervase Markham [:gerv]]

> So this feature works now, but the pref is set wrong?

The wrong pref is set. There's a subtle difference :-) Even if we are currently 
Doing The Right Thing, which is possible, this bug should remain open to fix 
that problem.

Gerv

Comment 8

[Gervase Markham [:gerv]]

It amazes me that this could happen. Did the implementor not do _any_ testing?

As far as I can tell, this stuff was checked in by Gagan in bug 1582, but
there's no patch attached and no indication of who reviewed it. <sigh>. The
evidence in the C++ code disappeared with the HTTP rearch.

Patch on its way. Looking for r= and rs=.

Once we've fixed the misnamed pref, we will then check whether it actually works
:-)

Gerv

Comment 9

[Gervase Markham [:gerv]]

Attachment: Patch v.1

Comment 10

[Bradley Baetz (:bbaetz)]

r=bbaetz

The actual problem isn't really the misnamed pref, since the code (quite
correctly) doesn't error out if the pref isn't found, and manually setting the
correct pref works fine. Its just a documentation issue - technically, nothing
is broken. However, someone else just reported this, so it would be nice to get
into the branch, if possible. ccing darin for an rs=

IIRC, pav fixed the image referrer thing some time in the last month, and I just
tested now that it works.

Comment 11

[Bradley Baetz (:bbaetz)]

*** Bug 83888 has been marked as a duplicate of this bug. ***

Comment 12

[Darin Fisher]

sr=darin

Comment 13

[Asa Dotzler [:asa]]

a= [e-mail] for checkin to the trunk.
(on behalf of drivers)

Comment 14

[Gervase Markham [:gerv]]

Checked in on the trunk.

Gerv

Comment 15

[John Unruh]

Verified fixed.

Comment 16

[benc]

John: what steps did you use to verify this?

Comment 17

[John Unruh]

I used the test URL at the top of the page.