From lwvanels Thu Feb  7 10:35:18 1991
To: kcunning@ATHENA.MIT.EDU
In-reply-to: kcunning@ATHENA.MIT.EDU's message of Wed, 06 Feb 91 17:57:46 EST <9102062257.AA16433@e40-371-1.MIT.EDU>
Subject: Thoughts about tracking OLH use 

Hi Kevin-
  I think it wouldn't be violating privacy too much to only collect the
userid and additionally to set the statistics file's permissions to be such
that the general public couldn't read it.  The number of unique users is
indeed something worth knowing (plus, since the pre-login use will have the
uid 0, it will also be possible to track how much use that is getting).
Here's specifically the information I propose collecting:

There will be two files used for collection of statistics- one that just
serves as a counter for the number of people who want to use it, and another
that will hold the statistics collected.

When a user starts up OLH, it will lock the counter file, read the value out
of it, increment the counter, write it back, and unlock the file.  This
gives us a simply count of the number of times OLH is being used, as well as
providing a unique "session id" to each run of OLH.

Then, in the main file, it will write

	session_id STARTUP time uid type

where "type" is either X or TTY.

Whenever a module is viewed, it would log into the main file the line

	session_id MODULE time node-id

These are both fairly simple changes to make, and they could be installed
and ready to go in a day or two (at the same time the new curses version is
deployed?)

When there's enough data, it would be easy to process using S- I've done a
fair amount of work with S when I did the usage statistics for OLC; these
staticstics are much simpler than that.

	-Lucien