36658 - [FEATURE][MEDIAMETRIX] Win32: need to enable MediaMetrix to count Mozilla usage

Status: VERIFIED FIXED

(SeaMonkey :: UI Design, defect, P3)

Description

[ekrock's old account (dead)]

To count usage of the Mozilla browser, on Windows, MediaMetrix needs to be able 
to track each URL the user visits. (Finished URLs only; not 
character-by-character keying in of a URL.)

For skinnability, we won't revert to use of a native Windows URL bar.

Appears that the easiest way to enable tracking would be able to create a 
hidden, invisible-to-the-user native Windows edit control that just duplicates 
the contents of our URL bar on a URL-by-URL basis.

Note that if we add this hidden control, MM doesn't need the DDE enablement 
mentioned in the below email as a possible need.

Here's the background info:

-----
Hello Chris and Mike,   this is a follow-up to my telephone conversation with 
Chris last week.  You may or   may not be familiar with Media Metrix but we are 
in the business of Internet audience measurement and   provide information 
services, including our rankings of top Web sites.    Media Metrix gathers its 
information   using a world wide panel of participants who volunteer to run our 
proprietary software (the meter) developed   by my technology team.   The mmxi 
meter measures page views, durations, etc at the browser and transmits   back to 
mmxi servers in real time. 

Teri Martola originally suggested I contact you regarding questions about the 
rollout of Netscape 6.0 and to review my proposal to Chris last week regarding 
our measurement of content viewed over Netscape technology.I would also like to 
establish an on going technical contact with product management/technology staff 
at Netscape with a goal toward evolving our metering technologies to most 
accurately measure Netscape's newest browser technologies - including the Gecko 
engine. Evolving our data capture technologies in conjunction with Netscape will 
allow Media Metrix to provide the industry the most accurate Web usage reporting 
and put accurate usage of Netscape technologies, regardless of platform,in the 
best possible light.. 

Our meters today run on Windows and Mac platforms.We use a combination of four 
techniques to detect web page viewingvia a browser or proprietary application 
connected to the Internet. First we have implemented a network device driver 
that allows us to determine files that have been downloaded to the computer via 
tcp/ip and standard Internet protocols.The driver allows us to capture web pages 
and streams downloaded to the PC, independent of the running application. Next, 
if available, we use the DDE protocol to capture URLS downloaded specifically to 
the browser. Finally we use a Windows hook to inspect the contents of the URL 
Address Window as the downloaded page appears on the display of the users 
browser (we also have the capability to establish a local proxy residing between 
the browser and the web server).Collectively these techniques allow the mmxi 
meter and data collection system to accurately measure a page as it is 
downloaded to the PC and displayed on the screen. 

We have been testing our meter technology with the new release Netscape 6.0 and 
noticed that Netscape has decided to forego the use of standard Windows edit 
controls in favor or proprietary controls.Additionally we note that the use of 
the DDE protocol has been removed from the browser. In order to provide the most 
accurate meausures of pages viewed we would like to explore the possibility of 
Netscape reinstating the use of a standard Windows Edit control in the URL 
Address Window of the browser.The existence of the standard edit control allows 
our service to provide a more accurate measure \of the page as it appears on 
display.I realize that you are probably trying to keep the code base platform 
neutral and the user interface consistent, thus if this is not possible, a best 
second alternative is to establish a secondary invisible Windows Edit Control 
that tracks the contents of the page currently in view.This edit control does 
not require the visibility bits be set, thus it will not interfere with the user 
interface of the browser. 

We have successfully worked with several companies to implement this technique 
and will certainly forward any further information as required. 

Moving forward, MMXI would definitely like to investigate using the Gecko engine 
to accurately measure Netscape on both Windows and non-PC based platforms.I 
would be very interested in finding out further plans Netscape has in developing 
and testing the Gecko engine on Internet appliances and set top boxes. It would 
certainly be great for the industry to provide accurate 3rd party measurement of 
Internet usage on all technologies andI believe we should be working together to 
make sure we have the best possible measurement solutions available. 

I look forward to your thoughts and recommendation for moving forward.. Would 
appreciate if we can have a followup discussion this week at your convenience. 
Please let me know your availability. 

Best Regards, 

George Ziegler
VP Technology Media Metrix World Wide
[phone]
[e-mail]
-----

Comment 1

[ekrock's old account (dead)]

Setting to M16 and marking nsbeta2, although there's no guarantee that we'll 
actually get this.

Also, note that MM will send us more info by email (which I'll attach to this 
report) about how to enable on Win32 for tracking. They'll also send info on how 
to enable on Mac for tracking (which I'll enter into a separate bug report).

Comment 2

[ekrock's old account (dead)]

Adding [MEDIAMETRIX] tag to Summary to enable easy tracking of all Media Metrix 
related bugs. Because this bug is Confidential, you must email it to 
[e-mail] if you want info from him.  Note related bug 36840 to enable 
tracking on the Mac via URL Echo AppleEvent.

More info from George:
-----
I would like to provide some additional background and detail information
regarding our near term request for NS6.0 enhancement.

The logic in our meter searches for edit controls within the active
application on the desktop.  When an edit control is found, the string
within the edit control is captured and validated for a URL string.  We
will search for edit controls in the top level/active application only and
only within the parent window of the application (the edit control can not
be embedded in a child window).  To identify a string as a URL, the string
must be prefixed with one of the following prefixes: http://, https://,
ftp://, file://, gopher://, telnet:// or news://.    If a url is entered
without the prefix, the browser normally prepends the http:// prefix (i.e.
www.aol.com would be displayed as http://www.aol.com).    We are requesting
that you place a non-visible windows edit control in the browser that
mirrors the content of the URL Address control.  The only requirement is
that this control be in the parent window with the WS_VISIBLE style turned
OFF.  As long as this control mirrors the URL control and has  the proper
prefix, the url information will be properly captured.

We use the information captured from the control so as to fine tune the
information captured on the screen, including pages that are viewed from
the cache.  When the page URL is in the URL Address window our meter
"thinks" the referring page is in view and allows us to accurately
accumulate time to the viewing.

I hope this is sufficient background and detail..   Please feel free to
contact me if any of the above is unclear.  We will be most happy to test
all changes once a decision is made to proceed forward.
Gz

George Ziegler

Comment 3

[ekrock's old account (dead)]

Phil Question:
The thing I'm not following is that these enhancements seem designed to  measure 
which web pages are viewed by N6 users, and for how long. Eric's  original 
request to us in the engineering group made it sound as though  the enhancements 
were required in order to compute market share data for  N6. Is that the case, 
or have I misunderstood?

Geroge Answer:
Our meter not only measures the page and page time, but it also 
associates which browser and which browser version has presented the page.    
Yes the requested change will allow us to better compute pages and page viewing 
time,  but in addition it will allow us to associate these metrics to the 
underlying browser version,  thus giving us a better picture of how the browser 
has been distributed across the panel.    Does this help?
Gz

Comment 4

[Phil Peterson]

-> don, cc phil

Comment 5

[leger]

Putting on [nsbeta2+] radar for beta2 fix.

Comment 6

[sairuh (rarely reading bugmail)]

shrir, would this fall under your plug-in/addon app component, at all?

Comment 7

[Phil Peterson]

This is a feature, so the + expires on 5/16

Comment 8

[ekrock's old account (dead)]

[pasting in email from jesse burris of MM; have emailed copies of the bugs]

Hi Eric,

I work at Media Metrix, and have been "offered up" as a resource here, to fix 
the bug reported by George.  I noticed that your e-mail to George mentioned two 
bugs.  Who can I contact to get copies of those reports? My Bugzilla search 
didn't find anything (my e-mail is [e-mail]).

I've successfully downloaded and built Mozilla (NS6 Browser). I've also 
identified, what I believe to be, the module (gkwidget.dll), and class 
(nsWindow) in question.  One instance of that class contains the "http..." 
string we're looking for.

There are, naturally, a few ways to add our invisible editbox to track the 
"http.." string.

Who can I contact to discuss a design?  From the Mozilla org's Module Owners 
document, it looks like Kevin McCluskey is the owner of the module in question.

I'm looking forward to working with the Mozilla organization, and look forward 
to your reply.

/jesse [e-mail] 516-507-3310

Comment 9

[don]

Need to find an owner to take the patch ...

Comment 10

[don]

Move to M19.

Comment 11

[leger]

Need to get more info on impact to not putting this into product.

Comment 12

[leger]

ccing hamerly.

Comment 13

[leger]

Putting on [nsbeta2+] radar for beta2 fix. 

Comment 14

[don]

OK, Bill says this is ready to go in now.

Comment 15

[ekrock's old account (dead)]

What's our status on getting this fix checked in? Are we still on track for
getting this in for nsbeta2?  Thanks to all for your hard work & efficient
coordination with external contributors!!!

Comment 16

[Bill Law]

Code has now been activated (should show up in tomorrow's build).  Jesse Burris 
at Media Metrix will let me know if they see any problems with it.

Comment 17

[sairuh (rarely reading bugmail)]

rubberstamp vrfy, as i'm not sure how we'd verify this from our qa perspective.

Comment 18

[Todd Pringle]

George Ziegler of MediaMetrix has given the okay to remove the Netscape
Confidential designation from this bug.

Comment 19

[Phil Peterson]

opening the bug