"Bryan Ax" <[EMAIL PROTECTED]> wrote:

> You wrote:
>
>> By the way, if your logon information is being tracked in a
>> cookie, then you must be checking that cookie within the page
>> access code. I presume that code uses some sort of timeout
>> mechanism? If so, then that's the place to count your sessions,
>> not in an after the fact log analysis.
>
> Do you have an example of counting the sessions in such a manner?

You mentioned ASP earlier in the thread. If you are using the ASP
Session_OnStart construct to initiate the login process, then you simpley
add a couple of lines to the Session_OnEnd function to log the username,
start time and end time. ASP calls Session_OnEnd automatically after a
specified timeout. If you're using some other mechanism, the basic
approach is the same - have whatever code generates the cookie keep track
of "sessions".

(If a user who leaves his or her browser open for a week at a time doesn't
have to log back in to your site, because you don't time them out, then
it's a bit pointless to talk about "sessions".

> I've
> done it this way before, however, I'd like to use something that's
> standardized.

If there was a standard, then you wouldn't be looking for advice here.
That's not meant to be a flippant response, but there isn't a standard way
of doing what you want to do, because the data isn't reliable enough to
support it.

> As for looking at 1000 reports - the purpose of this project is to
> see if different types of "users" use the site in different capacities,
> and how much they use it. So, eventually, SAS or some other stats
> package will have to look at all this data in order to determine if
> different types of users "used" the site more or less. How they want
> to determine use I don't know. I've warned them of the inaccuracies
> of "time spent online", however, they still may choose to use that.
> If so, then SAS will need to correlate all the sessions to determine
> if "nonsmokers", for instance, use the site more than "smokers" for
> accessing health-related information. Or diabetics, non-diabetics,
> females, males, or all types of lovely variables.
>
> Even if I setup the cookie as you describe, I still don't see any
> time-based reports in analog. Are they not even possible?

http://www.analog.cx/docs/webworks.html

Analog does time reports, but not what you call "time reports". In the
non-time reports, such as the User report, Analog will list the first and
last time of access (USERCOLS ED). If you generate a report against unique
"session-ids", this will give you some of what you want.

Aengus

+------------------------------------------------------------------------
|  This is the analog-help mailing list. To unsubscribe from this
|  mailing list, go to
|    http://lists.isite.net/listgate/analog-help/unsubscribe.html
|
|  List archives are available at
|    http://www.mail-archive.com/[email protected]/
|    http://lists.isite.net/listgate/analog-help/archives/
|    http://www.tallylist.com/archives/index.cfm/mlist.7
+------------------------------------------------------------------------

Reply via email to