Tom's Guide | Tom's Hardware | Tom's Games
 |
 |
 |
Dear Computing.net admins,
please remove the "Vary:*" header from your server configuration, this header prevents the caching of pages in Internet explorer, which means that the visitors have to load the pages again and again, since the cache is not used.
I removed the header using the Proxomitron proxy I am using to filter the web so that I can cache the computing.net pages, but not everybody uses proxies to access the web and I am sure the modem users would really appreciate it, since their connections are so slow that loading of pages takes much time.
Thanx
Hello,
I'm not sure what you are talking about. Memory caching works fine with my installation of Internet Explorer. If I click the back button, the page will not refresh.
If you are talking about disk caching, I don't think that is a good idea for this site. Otherwise, people might not be able to see if their questions have been answered very easily.
Justin
0 
No, I mean the normal caching, like you hit the back button in IE and you arrive directly at the previous page, because normally IE just takes the contents from the cache, but on your site it never did, so I looked at the headers and traced the problem to the "vary:*" header. There is much discussion about this header, you can learn about it searching on google. The problem with this header is that it forbids the browser to cache the page, forcing it to reload it every time. It may be useful sometimes, but definitely not on this page, since after disabling it with proxomitron the page still works fine, while caching started working.
This is the unfiltered reply that your server sends to the browser:
HTTP/1.1 200 OK
Date: Wed, 22 Oct 2003 02:34:39 GMT
Server: Apache
Vary: *
Content-Type: text/html
Connection: CloseSee the "vary" part? It prevents the caching of pages. It should be disabled.
Check this out:
W3c on the vary header:
http://www.w3.org/Protocols/rfc2616/rfc2616-sec14.html#14.44 (bottom of the page)
and so on
0
Ah, sorry, the post got defaced during posting, here go the links again:
http://lists.w3.org/Archives/Public/ietf-http-wg/2002AprJun/0046.html
http://archive.apache.org/gnats/4118
http://archivist.incutio.com/viewlist/css-discuss/30707
W3c on the vary header:
http://www.w3.org/Protocols/rfc2616/rfc2616-sec14.html#14.44 (bottom of the page)
and so on
0
I haven't had any problems with caching... I use Mozilla though, not IE, so it could be different for me...
0
Yes, this primarily affects the IE users, although Netscape 7 also doesn't cache computing.net pages, but always loads them anew. I am surprised that in your case caching works, because both mozilla and netscape are based on the same technology, in fact, netscape is just mozilla with a slightly different gui, the core is the same.
Opera has no problems with this, but only because that browser never sticks to any standards, but just does what it wants. This "vary" behaviour is in fact no bug, so IE and netscape do the right thing by not caching such pages, they stick to the standards. It's just that the forceful prevention of caching is not necessary on this page. Even if a visitor gets an outdated page from cache, all he has to do is to hit the refresh button to get the newest version, there is no need to force the browser to get the newest version from the server.
Considered the amount of people that come here the removal of this header will increase the server performance, since in most cases cached pages will be sufficient and the server won't get that many requests.
0
PHP,
I'm not sure what you are talking about. I visit clients that have 56K connections and I will go to Computing.Net while I am at said clients site. I can hit the back button and I have no problem going back quickly as well as the page refreshing itself if a new message has been posted or a reply to a message.
There is no slow down at all. I do this all the time from different client locations. Maybe I am missing something.
KTTD
0
@Kevin,
as I said, it doesn't affect all browsers. I don't know what browser you are using. Maybe with your browser it has no effect. If you say that you can go back quickly, so can I, lol (I'm on dsl), but that's not the point, the question is whether the browser loads the page every time anew, avoiding the cache.
You can use any sniffer - either Proxomitron (load the log window), or httpwatch (look at the received headers), or any other http sniffer, it will show "Vary *" when you load computing.net in a browser --> that's exactly the problem with caching, at least for the IE browser.
One can instruct the apache server (and that's the one used on this site) Not to send the vary header based on useragent, document type etc. But with this type one should disable the header altogether, at least I can't see any necessity in using it here.
0
btw, here's another good page that explains the problems with the vary header and internet explorer:
http://lists.over.net/pipermail/mod_gzip/2002-December/006838.html
0
PHP,
I am talking about IE and using many different versions of IE. Most folks that use 56K don't have the newer updated IE.
As I stated this is from many different client sites that have dial up access. Just today I did the same thing. No problem at all with the page loading when I hit the back key.
I do this all the time to just check on the site when I am away from home and never have had a problem.
KTTD
0
Well, if it works for you, fine, but doesn't work for me, and this is certainly not a problem with my computer, since I visit hundreds of sites every day, and till now I had caching problems only with 3 sites, and looking at the headers it turned out that all three are sending the vary header. And if you follow the link I gave above and read the article, you'll see that such a problem indeed exists. Even the apache page itself gives some info on this, and provides a solution, which was SPECIFICALLY developed for exactly this problem:
add to httpd.conf (apache 1.36 or later):
BrowserMatch "MSIE" force-no-vary
or for specific versions:
BrowserMatch "MSIE 4\.0" force-no-vary
BrowserMatch "MSIE 5\." force-no-vary
BrowserMatch "MSIE 6\." force-no-vary
etc.However, this won't help if a visitor obfuscates his user agent.
0
I am running IE version 6, SP1; and I am on a cable modem connection and have never had any difficulties with the above mentioned problems.......just thought I'd add the info to the thread. Tommyo
0
php
If there is a little blue 'new posts' Icon on a Forum List and I click on it and read the Post and then press the Back Button - the little blue 'thingy' is still there - suggesting that the page was taken from the cache
If the Page was re-loaded then the little blue 'thingy' would disappear wouldn't it ??
If I want to make sure that the Forum List is absolutley up to date - to the minute - then I press Refresh
I can't see what the problem is supposed to be ??
Lesley
0
[quote]
If there is a little blue 'new posts' Icon on a Forum List and I click on it and read the Post and then press the Back Button - the little blue 'thingy' is still there - suggesting that the page was taken from the cacheIf the Page was re-loaded then the little blue 'thingy' would disappear wouldn't it ??
If I want to make sure that the Forum List is absolutley up to date - to the minute - then I press Refresh
I can't see what the problem is supposed to be ??
[/quote]
yes, this is absolutely correct, and your description suggests that your browser indeed caches the pages. So you are lucky, but in many cases, like in mine it can happen that the mentioned header will prevent the caching, as it happens in my case. Then IE basically always "refreshes" the page, not just when I hit the refresh button. I don't quite understand the criteria for this behaviour, since I also have the latest version of IE with all updates and everything, and still I've got this problem. Still, my point is that if such a header is unnecessary and brings no advantages, then it should be disabled, even if some people might have no problems with caching. It's really simple - if it's useless for 50% of visitors and damaging for another 50%, then the best solution is to remove it altogether
0
I am presuming it like the feature on IIS for http keepalives. The pages stay in cache so if the user asks for that page again, it does not have to go out and get the page again because it is in cache. The end result being less "new" requests being made to the server so it increases performance on the webserver and the user is served their pages slightly faster.
0  |
 |
 |
This post is quite old and has been locked from receiving new replies. Please create a new posting instead.
| Ads by Google |
