Re:st: Stata agent string for net install

[ariley@stata.com (Alan Riley)]

Brendan Halpin (brendan.halpin@ul.ie) asked about a set of four
numbers Stata includes in its "browser id" string that it reports
when it accesses files over the web:

> Does anyone happen to know what the four numbers in the agent string
> that Stata sends when making a HTTP connection to do a -net install-?
> 
> That is, when you do a -net install- Stata identifies itself as a
> browser to the web server. Older versions may simply say "Stata/9.2
> (Windows)" but more recent ones give more info, e.g.:
> 
> Stata/IC 11.1 (418.11.1.687) on Windows NT 5.0
> 
> What does the 418.11.1.687 mean (I've made up the numbers here in case
> they actually reveal information about people)? Is it unique to the
> Stata installation? Can I use it, for instance, to enumerate unique
> downloads (better than using IP addresses)?

Unfortunately for Brendan's purposes, but fortunately for our users,
this number does NOT uniquely identify Stata installations nor Stata
users.

This string is known as a browser id string, or, more formally,
a "User Agent ID".  It is how web browsers, and applications
like Stata which access the web, tell web servers about themselves.

Here is what one web browser might report:

   Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.1; SV1)

The parts of this that matter to most people are "MSIE 6.0", which
tells you that the browser being used is Microsoft Internet Explorer
6.0, and "Windows NT 5.1", which tells you that the browser is running
on Windows XP 32-bit.  Yes, Microsoft has called every version
of Windows since the original Windows NT 3.1 "Windows NT".  Windows 8
will be "Windows NT 6.2".  (By the way, hopefully the above browser id
string didn't come from any of you.  Internet Explorer 6 is a legacy
browser full of security holes.  Even Microsoft wants you to stop using
it: http://www.ie6countdown.com/ )

Brendan made up what he reported above.  Here is a real Stata example
along with an explanation of it:

    Stata/MP 12.1 (521.12.1.887) on Windows NT 6.1

Let's break that down.

    Stata/MP 12.1       (obvious)
    521.12.1.887        (see below)
    on Windows NT 6.1   (running on Windows 7; I'll explain below
                         why 'Windows NT 6.1' is reported)

Looking closer at the four numbers in the middle, we see they are
really three:

   521     (an encoded version of the Stata flavor and operating system)
   12.1    (the Stata version)
   887     (our internal build number)

The 12.1 in the middle has an obvious meaning.

The 887 on the end is our internal update build number.  887 corresponds
to the current update of Stata 12.1 which users have in their hands now.
FYI, the previous update of Stata 12.1 which users saw was 884, and
internally at StataCorp we had 885 and 886 which users never saw as we
were working toward 887.

The 521 at the beginning contains two pieces of information.  The first
is the flavor.  5 corresponds to Stata/MP.  (4 is Stata/SE, 3 is
Stata/IC, and 2 is Small Stata.)  21 is 64-bit Windows.  There are
different numbers for other operating systems.

All users using the latest update of Stata/MP 12.1 on 64-bit
Windows 7 would have the same string.  So, Brendan cannot use
this string to uniquely identify hits to his website from different
Stata users.  The IP address would be a better choice for this even
though it isn't perfect either.  Even if Brendan sees two different
Stata User Agent IDs from the same IP address, he has no way of
knowing whether they are from two different people or if they are
merely a single person who has Stata on, say, both a PC desktop and a
Mac laptop.

In any case, Stata reports this string for two reasons.  First,
any application which accesses files from a web server is supposed
to report such a string.  Second, in the case of a technical support
question related to Stata accessing files over the web, this string
gives our Technical Services group some very useful information about
Stata and the operating system on which it is running.

A quick caveat: although we have no plans to change it right now, we
don't promise to keep the User Agent ID reported by Stata in the same
format in the future, nor will the pieces of the quartet of numbers in
the middle of it necessarily mean the same thing in the future.


--Alan
(ariley@stata.com)

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/