Welcome to WebmasterWorld Guest from 126.96.36.199
Forum Moderators: open
Take a few seconds and your will figure out why.
But I never expect that syntax to works! Expert please explain.
Google has a selection of language sites eg
For French, with "es" for Spanish, "eng" for English, etc
The "English" Google you have found with the trailing "dot" is not part og this group, but is not the main Google site either (that does not use the word "English" on it)
So, someone can explain technically how the . works?
very odd.. Your browser most likely tidies up the url the best it can before resolving the name to an actual machine but probably sends the . thought when it requests the page, so what site is returned is a matter of which virual host the server decides to return.. Perhaps the 'English' version of google is simply first/last/default in the list of vhosts?
(I can't get our (apache) servers to accept the trailing . it just resolves the address without it.) But google are using their own server GWS/2.0 so this is all just a big guess ;)
In disk subdirectories, "." and ".." have special meanings. ".." is a shortcut to the parent directory, and "." is a shortcut for the current directory.
Those familiar with DOS should recognize the command "CD .." which will take you up a directory level.
When you use just one period, you are asking for the current directory. So when you ask for this web page -
You are simply asking for the current directory that you are already on, so you get the home page.
Google seems to see this as a different URL, thus the difference in PageRank.
Depending on how a site's hosting is set-up, you may be able to use the ".." option in a url as well.
For example, try -
rather than give you a page in their windows subdiectory, you will find that this displays their home page!
Anyhoo.. in the url the dot comes before the first slash:
google.com./ rather than google.com/.
Still not convinced there's anymore truth in my explanation though! ;)
For example, at Microsoft (which is likely hosted on Windows), this works just fine -
Google doesn't like this, but then they aren't hosting on Windows either ;)
Also, remember that using a period rather than a slash at the start of a directory path in Windows is perfectly OK.
means to start from the current directory, and
means to start from the root directory
It just so happens that on a web site, they both usually point to the same thing!
[edited by: aspdesigner at 1:33 pm (utc) on Feb. 26, 2003]
The Domain Name System is a tree like system. The trailing dot is the root of that domain tree.
See RFC1034 DOMAIN NAMES - CONCEPTS AND FACILITIES [faqs.org] for more infos on DNS.
is entirely different from
. Only in the first example does the dot belong to the authority section of the URL. In the second example any UA will connect to google.com and request the resource
. In the third example it will indeed request
will just be removed from the URL (RFC2396 - Uniform Resource Identifiers (URI): Generic Syntax [faqs.org]).
[edited by: andreasfriedrich at 1:33 pm (utc) on Feb. 26, 2003]
Let me show you another way.. Google's 'Long IP' is.. 3639554917
This means you can access Google via [3639554917...] which results in the same page as the one with the extra dot :) In this case, however, it gets no pagerank at all ;-)
Long IPs are often used by spammers to disguise their servers.. but they're legitimate too.
Since their is nothing after the "." it will simply ask the root servers were "com" is. Then ask "com" where "google" is then ask "google" where "www" is then go there.
That's not right
[google.com...] which is plain Google
is a completely different page to
[google.com....] which is Google "English"
cornwall is quite correct, the "." is being passed to and resolved by the web server, as these two URLs display DIFFERENT pages!
Yes and no. Of course the trailing dot (as part of the host name) will be passed to the web server in the host field of the request header. And the web server does use that field to do something like resolving.
Both google.com and google.com. resolve to the same ip address (188.8.131.52 or 184.108.40.206) since they are the same. This is probably the result of some round robin DNS load balancing.
When you request a page the browser will connect to the web server and then request a certain URL. The resolving of the authority section of a URL is done on the client side or by the client sending a request to its domain´s NS which will then query other name servers when it is not able to resolve the domain name itself.
>>as these two URLs display DIFFERENT pages!
The effect you are seeing is Google´s geo-targeting at work.
in DNS I beleive (but am not absolutly sure) that the trailing "." is an understood. That is the root of DNS and it's what the rootservers control. Under that their is "com" "org" "net" and all the other TLD (top level domains) and it goes down from there.
So there is *NO* difference between "www.google.com." and "www.google.com" as far as DNS goes. As a matter of fact if you do "dig www.google.com +trace" it will add the trailing "." in the results because it's an understood.
My guess on the PR difference is that googles PR has nothing to do with DNS. It's a string match, in that context "www.google.com"!= "www.google.com." and theirfor as far as PR goes they are different sites.
None the less it is quite interesting to see the different PR's
You tyle google.com. on the browser. Your computer try to resolve the without period version and get the IP. (What I say the DNS server threat both version the same).
But THEN due to virtual hosting. When the browser requesting data from the IP, the browser do send to the web server the domain name which is google.com. to know which web sites to display. (This are all about domain name virtual hosting, same ip, hosting different web sites).
So I guess google.com. will return the default site by the web server. Which is the ENGLISH version of google. I fully agree with 'yetanotheruser'.
For HuhuFruFru, the . must be appear in the href attribute of the link.
I have a DNS proxy installed on my machine. It allows me to view all DNS requests made by my browser.
When I tried to view the web page -
My browser makes a DNS resolution request for this domain -
with no period at the end. The period is not included in the DNS request (whether serviced locally from cache or sent to an external DNS Server), and it is not used to determine the host address.
The extra "." is only included as part of the page request that is sent to the web server after the host address has been resolved.
daisho, I believe you may be correct. It is possible that the browser may be stripping-off the "." as redundant, thus it would never issue a DNS request for "www.google.com." as it would interpret it simply as a request for "www.google.com"
As it is never seen by the DNS, it would be ignored except is cases where the server is specifically examining the entire URL, such as what might happen with virtual web servers that share the same IP.
Google appears to be using a kind of "virtual server" approach. It appears that Google is using domain name DNS resolution only to direct you to the appropriate data center, but using a virtual server approach to direct you to the proper language, by examining what URL you typed. For example, www.google.de, www.google.fr and www.google.co.uk all appear to be CNAME aliases of www.google.com, they all would have the same IP address!
It appears that the web page displayed when you try www.google.com. (with the extra period) is actually the home page for "www.google.us"
With regards to the difference in PR, this is not unexpected, as Google is already known to do this in other circumstances. For example, try a search for -
This is why it is a good idea to make sure that all of your inbound links are consistent (i.e. - www.mysite.com vs. mysite.com) in order to avoid splitting you PR into two "different" sites.
The problem is that the server never sees the entire URL. In the request the browsers will only give the absolute path. To allow for name based virtual hosting the hostname is included in the host header field. See Pointing multiple domain names to main site without mirrors - How to do this without hosting them separately and using 301s? [webmasterworld.com] for an in-depth discussion on virtual name based hosting.
The problem is that the server never sees the entire URL. In the request the browsers will only give the absolute path.
Perhaps you need better hosting?
With regards to the other thread you mentioned, I took a look at it, but it seemed to be mostly about domain aliasing ("parking") rather than virtual hosting.
However, I would expect that Google implemented this using something much more sophisticated than the simplistic approach described there. My intent was simply to compare their implementation of the foreign google sites to virtual hosting only as to the end result - serving-up different content from the same IP based on the domain name in the URL.
What is going on here?