How do I create a log file that tells me who visited my page?

137 views
Skip to first unread message

nowi...@sciborg.uwaterloo.ca

unread,
Jul 11, 1995, 3:00:00 AM7/11/95
to
How do I create a log file that tells me who visited my page?

--
Jack Nowinski
E-Mail: nowi...@sciborg.uwaterloo.ca


Patrick Lee

unread,
Jul 11, 1995, 3:00:00 AM7/11/95
to
nowi...@sciborg.uwaterloo.ca () wrote:

> How do I create a log file that tells me who visited my page?

You cannot. There is absolutely no fool proof way to tell WHO has read
your pages -- w/o using authentication. The most you can get is WHERE
an access came from but that gives you zero information because there
can be anywhere between one and thousands of users on the other end.

--
Patrick Lee Stuyvesant H.S. Alumni Assoc.
Internet: pat...@panix.com Internet: pat...@panix.com
Web: http://www.panix.com/~patlee Web: http://www.panix.com/~stuy


Ingo T. Storm

unread,
Jul 12, 1995, 3:00:00 AM7/12/95
to
nowi...@sciborg.uwaterloo.ca () wrote:
>How do I create a log file that tells me who visited my page?

Doesn't your http daemon do that for you? This is of course a log that lists every
single access in one line. You should then use a tool like MUSAGE to summarize the
accesses. You can have a look at the results on <URL:http://shasta.ct.ix.de:8080/usage/husage.htm">
There's also a link to the origin of MUSAGE on that page.

Cheers,
Ingo

-----------------------------------------------------------
Ingo T. Storm eMail i...@ct.ix.de
Redaktion c't CIS 72662,3142
Verlag Heinz Heise
Helstorfer Str. 7 voice +49 (511) 5232-318
D-30625 Hannover, GERMANY fax +49 (511) 5352-417
-----------------------------------------------------------
This sentence no verb.

Matt Watson

unread,
Jul 12, 1995, 3:00:00 AM7/12/95
to
pat...@panix.com (Patrick Lee) wrote:

:nowi...@sciborg.uwaterloo.ca () wrote:
:
:> How do I create a log file that tells me who visited my page?
:
:You cannot. There is absolutely no fool proof way to tell WHO has
read
:your pages -- w/o using authentication. The most you can get is
WHERE
:an access came from but that gives you zero information because
there
:can be anywhere between one and thousands of users on the other end.

Ok, then how does one set-up a way to see WHERE the visits came from?

Patrick Lee

unread,
Jul 13, 1995, 3:00:00 AM7/13/95
to
Matt Watson <wat...@jeflin.tju.edu> wrote:

> pat...@panix.com (Patrick Lee) wrote:
> :
> : You cannot. There is absolutely no fool proof way to tell WHO has read


> : your pages -- w/o using authentication. The most you can get is WHERE
> : an access came from but that gives you zero information because there
> : can be anywhere between one and thousands of users on the other end.
>
> Ok, then how does one set-up a way to see WHERE the visits came from?

It's set in the server configuration, which is server dependent. Ask
your sysadmin for the access log file or if you are the sysadmin, RTFM!

Rasmus Lerdorf

unread,
Jul 13, 1995, 3:00:00 AM7/13/95
to
nowi...@sciborg.uwaterloo.ca () writes:

>How do I create a log file that tells me who visited my page?

The PHP Tools can do this for you. Have a look at http://www.io.org/~rasmus
for an example.

--
Rasmus Lerdorf
ras...@io.org

Patrick Lee

unread,
Jul 14, 1995, 3:00:00 AM7/14/95
to
a5g...@rick.cs.ubc.ca (Wei-Yuen Tan) wrote:

> % You cannot. There is absolutely no fool proof way to tell WHO has read
> % your pages -- w/o using authentication.
>
> Actually, there is, provided the client machine is running identd. Check out
> RFC 1413, and then look at the bottom of my homepage (the lewd part). Of
> course, you can toss this out the window if you're not running identd.

How do you know you can TRUST identd? And how many systems out there do
you think is running identd? A few percentage, perhaps. identd is
useless for people coming from behind firewalls, PCs, Macs, Compuserve,
AOL, Prodigy, etc. These people probably account for the bulk of the
web users right now.

Rasmus Lerdorf

unread,
Jul 14, 1995, 3:00:00 AM7/14/95
to
FLA...@cernvm.cern.ch (Alan J Flavell) writes:
>This is a much more economical way (for you and for the network) than
>trying to turn every page into a script, just so that you can append
>some CGI environment stuff to a file of your own.

The above is misleading. It insinuates that a cgi program takes up
more network bandwidth than a straight HTML page. This is not the
case. If the page looks the same in both cases, the amount of
network traffic is actually exactly the same. The only reason one
might say that it is more "economical" to not generate HTML pages
from CGI scripts is that a CGI will take a bit more CPU on the server
machine than if the page didn't need to run a cgi.

--
Rasmus Lerdorf
ras...@io.org

Alan J Flavell

unread,
Jul 14, 1995, 3:00:00 AM7/14/95
to
In article <3u1ld3$9...@elmo.tju.edu>

Matt Watson <wat...@jeflin.tju.edu> writes:

>Ok, then how does one set-up a way to see WHERE the visits came from?

Most servers keep a log. Consult your webmaster, or, if you are the
webmaster, RTFM (or if all else fails, consult the appropriate
c.i.www.servers... group).


This is a much more economical way (for you and for the network) than
trying to turn every page into a script, just so that you can append
some CGI environment stuff to a file of your own.

However, do be aware that for a single access by a caching proxy, there
could be hundreds or thousands of accesses that you never see.

And the previous discussion was right - mostly you don't get told
_who_ did the access, and even if you are told, the information is
very unreliable.

Wei-Yuen Tan

unread,
Jul 14, 1995, 3:00:00 AM7/14/95
to
In <3tveji$r...@news.panix.com> pat...@panix.com (Patrick Lee) writes:

%> How do I create a log file that tells me who visited my page?

%You cannot. There is absolutely no fool proof way to tell WHO has read
%your pages -- w/o using authentication.

Actually, there is, provided the client machine is running identd. Check out
RFC 1413, and then look at the bottom of my homepage (the lewd part). Of
course, you can toss this out the window if you're not running identd.


--
** Wei-Yuen Tan monk...@pobox.com http://pobox.com/~monkeyboy **
#!/usr/local/bin/perl -s-- export-a-crypto-system sig, RSA in 3 lines PERL:
($k,$n)=@ARGV;$v=$w=1+length$n&~1;$v-=$d*2;$w-=$e*2;$m=unpack(H.$w,$m),$_ = `echo "16dio1[lm*ln%]sz\U$k $m $n\Esnsm[d2%Sa2/d0&lt;xsbd*ln%La1=z2]dsxxip "|dc`,s/\W//g,print pack(H.$v,0 x($v-length).$_)while read(STDIN,$m,$w/2)

Alan J Flavell

unread,
Jul 15, 1995, 3:00:00 AM7/15/95
to
In article <3u4d74$d...@nnrp.cs.ubc.ca>
a5g...@rick.cs.ubc.ca (Wei-Yuen Tan) writes:

>% There is absolutely no fool proof way to tell WHO has read

>%your pages -- w/o using authentication.
>
>Actually, there is, provided the client machine is running identd.

Please stop repeating this misinformation, as it's always possible
that some confused reader might believe you. The correct answer has
already been given several times on this thread _and_ on several
related threads that are running currently.

I can easily set up IDENTD to say that I am Mickey Mouse. All of
the other thousand users who visit that kiosk station will also
report that they are Mickey Mouse. And the same goes for the other
five kiosk stations in that cluster, if I choose to make it so, What
the quote says is perfectly correct. There is ABSOLUTELY NO FOOL PROOF
WAY. This has been correctly stated several times.

Alan J Flavell

unread,
Jul 15, 1995, 3:00:00 AM7/15/95
to
In article <3u5oli$r...@ionews.io.org>

ras...@io.org (Rasmus Lerdorf) writes:

>FLA...@cernvm.cern.ch (Alan J Flavell) writes:
>>This is a much more economical way (for you and for the network) than
>>trying to turn every page into a script, just so that you can append
>>some CGI environment stuff to a file of your own.
>
>The above is misleading. It insinuates that a cgi program takes up
>more network bandwidth than a straight HTML page.

Many cache proxy servers assume that the results of a CGI script
cannot be cached, so I do not believe I was wrong, or even "misleading",
in drawing attention to the network implications of turning every
plain web page into a script. One might not have resented your
reply quite so much, if you had asked for clarification of what was
meant, rather then boldly asserting that it wasn't so.


> If the page looks the same in both cases, the amount of
>network traffic is actually exactly the same.

The volume of traffic is the same, but if it has to come from the
distant host rather than from the local cache server, the implications
for the network are significant - especially on congested international
links.


> The only reason one
>might say that it is more "economical" to not generate HTML pages
>from CGI scripts is that a CGI will take a bit more CPU on the server

That might have been the only reason that you could think of, but it
isn't the only reason, as I have explained.

Sandra Steingart

unread,
Jul 15, 1995, 3:00:00 AM7/15/95
to nowi...@sciborg.uwaterloo.ca

Bill Graham

unread,
Jul 15, 1995, 3:00:00 AM7/15/95
to FLA...@cernvm.cern.ch
You indecate that you can easily set up IDENTD to indicate whatever you want. I am working on
my own back end security system and am looking for a way for the user to load their user-id and
password in Netscape variables so that this information will be sent as part of theheader every
time the user interacts with the Server. Could you please describe how this can be done.

Thanks in advance for any assistance.

Bill Graham

Martian

unread,
Jul 19, 1995, 3:00:00 AM7/19/95
to
ras...@io.org (Rasmus Lerdorf) writes:

++FLA...@cernvm.cern.ch (Alan J Flavell) writes:
++>This is a much more economical way (for you and for the network) than
++>trying to turn every page into a script, just so that you can append
++>some CGI environment stuff to a file of your own.

++The above is misleading. It insinuates that a cgi program takes up
++more network bandwidth than a straight HTML page. This is not the
++case. If the page looks the same in both cases, the amount of
++network traffic is actually exactly the same. The only reason one
++might say that it is more "economical" to not generate HTML pages
++from CGI scripts is that a CGI will take a bit more CPU on the server
++machine than if the page didn't need to run a cgi.

That is not quite true. An HTML page can be cached. The output of
a CGI-program is always `fresh', and hence cannot be cached.

Abigail

Reply all
Reply to author
Forward
0 new messages