Article View: alt.folklore.computers
Article #116008Re: Mars and the Internet
From: Keith M Ellis
Date: Mon, 14 Jul 1997 00:00
Date: Mon, 14 Jul 1997 00:00
46 lines
3045 bytes
3045 bytes
In alt.folklore.urban Abigail <abi...@fnx.com> wrote: > Foo! Basically, you cannot track the visitor through a site. HTTP is a > stateless protocol. You can try to get some results with cookies, but > that doesn't work if cookies are disabled (which can easily be done by > a proxy), if the user uses multiple browsers, if a browser is used by > various people (think public accessable terminals like in libraries and > universtities), or if there are proxies involved meaning that only > certain page visits register hits at the server. I don't understand this obsession with cookies. Cookies are irrelevant to this discussion. Most browsers report the referring URL to the server when linking to another page. Voila, you can track a visitor through a site. I do it all the time - all of my web pages trigger an SSI that logs the IP address/fqdn, date, browser type, and referring url for each access. I use this to look at my logs to track user movements through my web site. Since it's an SSI, it's entirely invisible to the user - no cgi calls in the source visible to the browser, no cookies, no java, no whatever. I know who visits my site (I know the ip address, at least), and I usually know the URL they linked from, because the http definition includes making this information available to the server. Period. Not a damn thing to do with cookies. Let's get over the cookie hysteria, in this regard, at least. The unmassaged server logs I've seen log all hits to everything - that's every single URL asked for in the HTML source (images, whatever). But it's not hard to filter out everything but page URL's. Which are they calling "hits?" Hopefully page hits. A counter program is counting page hits, or something that is functionaly equivalent. But people will often reload a page several times - for a variety of reasons. That inflates the count. More and more people are looking at the pages via a proxy server. That means that a lot of legitimate "hits" of a page are not being seen at all by the server. Given AOL, et al, I'd say that the latter more than makes up for the former, depending on how quickly a proxy expires its cache. I've personaly looked at the pathfinder page(s) about twenty times, most of those via my ISP's proxy. [least relevant froups trimmed from followups] -Keith "I'm watching" Ellis
Technical Details
Message-ID:
<5qc9i7$7j3$1@nntp-2.io.com>Path:
mbox-mo1.pugleaf.net!unknown!not-for-mailReferences:
<5ptll4$hnq@news1.panix.com> <5pup0j$pbo@winter.erols.com> <5q1s01$68n@news.inforamp.net> <5q21ur$9k5$1@turing.wins.uva.nl> <5q5cbi$t4u@sulawesi.lerc.nasa.gov> <ED8ts0.Ltp@nonexistent.com>Show Full Headers
Complete Article Headers
From: Keith M Ellis <kme...@io.com> Subject: Re: Mars and the Internet Date: 1997/07/14 Message-ID: <5qc9i7$7j3$1@nntp-2.io.com> X-Deja-AN: 256755857 References: <5ptll4$hnq@news1.panix.com> <5pup0j$pbo@winter.erols.com> <5q1s01$68n@news.inforamp.net> <5q21ur$9k5$1@turing.wins.uva.nl> <5q5cbi$t4u@sulawesi.lerc.nasa.gov> <ED8ts0.Ltp@nonexistent.com> Organization: International House of Pain Cakes X-URL: www.io.com/~kmellis Newsgroups: alt.folklore.urban,rec.org.mensa,sci.skeptic,sci.math,comp.misc,alt.folklore.science,alt.folklore.computers MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 7bit