Forum Moderators: open
I was wondering if anyone has started using it in their content and has noticed it used by any users, or if anyone has turned it on in their browsers and noticed it doing anything. I understand it's one of those deals that both parties have to start using, so it's unlikely that it's caught on enough to begin to work, but any experiences would be interesting.
Do any browsers other than Mozilla support this feature? I have a higher-than-usual proportion of alternative browser users, but even so Moz 1.2b isn't a blip on the radar.
On the other hand, those that rely on advertising could find the feature most profitable. Cram those links onto the home page and rest assured of a few dls per moz user - money in the pocket.
Do any browsers other than Mozilla support this feature?
The beta versions of the Mac browser iCab [icab.de] have had a pre-fetch feature for a while. When I was experimenting with iCab I used this feature a little, but it generated so much churning in the background that I left it off most of the time.
HTML 4.01 Specification [w3.org]
For instance, links defined by the LINK element may describe the position of a document within a series of documents. In the following excerpt, links within the document entitled "Chapter 5" point to the previous and next chapters:<HEAD>
...other head information...
<TITLE>Chapter 5</TITLE>
<LINK rel="prev" href="chapter4.html">
<LINK rel="next" href="chapter6.html">
</HEAD>The link type of the first link is "prev" and that of the second is "next" (two of several recognized link types). Links specified by LINK are not rendered with the document's contents, although user agents may render them in other ways (e.g., as navigation tools).
I've been using <link rel="next".... for about 3 years now and has been supported by IE4+ to my knowledge.
If use increases, this will also make web stats meaningless.
That would be very bad - throwing away one of the most valuable aspects there is in web marketing.
However, it certainly is a potential. Many businesses make only the most minimal use of their server logs - they just haven't got a handle on the value yet. But I've worked with businesses who've fine-tuned or even redirected their offline business plans based on what they've learned online.
I'd hate to see that potential be diminished - especially when broadband obviates the need for most preloading.
Do prefetched requests contain a Referer:header?
No, prefetched requests do not contain a HTTP Referer: header. This is because the user has not explicitly requested the prefetched document. This certainly may impact referrer tracking that is commonly used on many sites. For this reason, link prefetching may not be appropriate for all content.
It should be easy to filter prefetched requests.
<added>
user_pref("network.prefetch-next", false); on your user.js switches it off. </added>
<added2> Ouch. It's going to be impossible to determine if a prefetched pages was viewed or not... </added2>
[edited by: Duckula at 5:10 pm (utc) on Oct. 18, 2002]
We are considering adding UI for this preference; however, the overriding theory is that if link prefetching needs to be disabled then there must be something wrong with the implementation.
Looks like we'll have to get used to filtering out requests with no referrer in our stats. After all, how many users are going to edit their prefs.js? Heck, I consider myself a programmer before a web-anything, and I've never been motivated enough to look at prefs.js.
I certainly wouldn't take such a decision lightly, but Moz is out in left field putting an auto spider in surfers hands. I also agree that it is a user feature - we _need_ more user features in browsers today to handle the hostile web. Unfortunatly, this feature does nothing to make the browser easier to use with so many trashed out pages out there. It just causes server abuse and it's utility to users is marginal.
Any one with a counter opinion on why we should allow mozilla to cause such server abuse?
Any one with a counter opinion on why we should allow mozilla to cause such server abuse?
I'm not going to excuse such a design faux pas. But Brett, I'll keep using latest Mozilla builds, I'm going to keep coming here, I'm switching off that feature at user.js, and if you keep me out you're going to see a few odd useragent strings at your logs :)
Besides, there's no <link>ing in the <head> of this page...
They actually download secure information and cache it without users authorization or control unless the user has read that page and possibly can figure out what a prefs.js is.
That is wrong!
I hope this doesn't filter over to the next Netscape.
Bandwidth use by Mozilla users is up 30% the last day.
Perhaps because a new version of mozilla is out. I wouldn't attribute it to the prefetching necessarily. If you ban Mozilla, I'll have to change my useragent string, use Konqueror (or some other linux client), surf webmaster through google's cache, go back upstairs to IE on my mac, or just quit coming around. I can't imagine you wanting your users to have to do any one of those.
Perhaps Mozilla needs to change its useragent string when it's prefetching so you can refuse those queries, or it needs to obery robots.txt or something. That would be great.
I am rather excited about link prefetching, and I'd like to see it work. Perhaps I'll direct the mozilla developers to this thread so that they can see some of the discussion.
Any one with a counter opinion on why we should allow mozilla to cause such server abuse?
Are you sure you thought this one through carefully enough?
Mozilla will prefetch a page only when another page contains an explicit prefetch link to it. The user has no way of telling Mozilla "please prefetch all links on this page". It is all in the hands of the webmaster.
If anything, then 30% increase in Mozilla bandwidth use is a sign that more and more web savvy people are switching to decent and up to date client software. If you want to keep those folks away from here, then you know what to do... ;)
[edited by: bird at 10:11 pm (utc) on Oct. 19, 2002]
I'd hate to have to use IE. *ick* :)
https:// URLs can be prefetchedThey actually download secure information and cache it without users authorization or control unless the user has read that page and possibly can figure out what a prefs.js is.
This is not "secure information". This is information that was transmitted between the server and the client in a way that is supposed to prevent interception by a third party. If the server maintainer decides that it is ok to allow caching this information, why should that be a problem for the user?
Do prefetched requests contain a Referer:header?
Yes, prefetched requests include a HTTP Referer: header indicating the document from which the prefetching hint was extracted.
As a server admin, can I distinguish prefetch requests from normal requests?
Yes, we send the following header along with each prefetch request:X-moz: prefetch
