Not registered yet?
Register now! It is easy and done in 1 minute and gives you access to special discounts and much more!
By Haim Evgi
I would like to know the differences between these two types of URLs: relative URLs (for pictures, CSS files, JS files, etc.) and absolute URLs.
In addition, which one is better to use?
Assume we are creating a subsite whose files are in the folder http://site.ru/shop.
Link to home page
Link to the product page
Link from home page to product page
Link from product page to home page
Although relative URL look shorter than absolute one, but the absolute URLs are more preferable, since a link can be used unchanged on any page of site.
We have considered two extreme cases: "absolutely" absolute and "absolutely" relative URLs. But everything is relative in this world. This also applies to URLs. Every time you say about absolute URL, you should always specify relative to what.
Link to home page
Link to product page
Google recommends such URL. Now, however, it is generally considered that http:// and https:// are different sites.
I.e. relative to the root folder of the domain.
Link to home page
Link to product page
It is a good choice if all pages are within the same domain. When you move your site to another domain, you don't have to do a mass replacements of the domain name in the URLs.
The tag <base> specifies the base URL, which is automatically added to all relative links and anchors. The base tag does not affect absolute links. As a base URL we'll specify the home page: <base href="http://sites.ru/shop/">.
Link to home page
Link to product page
Now you can move your site not only to any domain, but in any subfolder. Just keep in mind that, although URLs look like relative, in fact they are absolute.
Especially pay attention to anchors. To navigate within the current page we have to write href="t-shirts/t-shirt-life-is-good/#comments" not href="#comments". The latter will throw on home page.
For internal links I use base-relative URLs (5). For external links and newsletters I use absolute URLs (1).
great answer. too bad it lingers too low on the page for people to see it. answers a lot of the comments on other answers.
If by absolute URLs you mean URLs including scheme (e.g. http / https) and the hostname (e.g. yourdomain.com) don't ever do that (for local resources) because it will be terrible to maintain and debug.
Let's say you have used absolute URL everywhere in your code like <img src="http://yourdomain.com/images/example.png">. Now what will happen when you are going to:
In the first example what will happen is that you will get warnings about unsafe content being requested on the page. Because all your URLs are hardcoded to use http(://yourdomain.com/images/example.png). And when running your pages over http the browser expects all resources to be loaded over https to prevent leaking of information.
In the second example when putting your site live from the test environment it would mean all resources are still pointing to your test domain instead of your live domain.
So to answer your question about whether to use absolute or relative URLs: always use relative URLs (for local resources).
First lets have a look at what the difference URLs are we can use:
In the examples below I assume the website is running from the following location on the server /var/www/mywebsite.
The above (absolute) URL tries to access the resource /var/www/website/images/example.png. This type of URL is something you would always want to avoid for requesting resources from your own website for reason outlined above. However it does have its place. For example if you have a website http://yourdomain.com and you want to request a resource from an external domain over http you should use this. E.g. https://externalsite.com/path/to/image.png.
What this type of URL does is use the current scheme of the page it is on. This means that you are on the page http://yourdomain.com and on that page is an image tag <img src="//yourdomain.com/images/example.png"> the URL of the image would resolve in http://yourdomain.com/images/example.png.
When you would have been on the page http**s**://yourdomain.com and on that page is an image tag <img src="//yourdomain.com/images/example.png"> the URL of the image would resolve in https://yourdomain.com/images/example.png.
This prevent loading resources over https when it is not needed and automatically makes sure the resource is requested over https when it is needed.
The above URL resolves in the same manner on the server side as the previous URL:
The above (absolute) URL tries to access the resource /var/www/website/images/example.png.
The above (absolute) URL tries to access the resource /var/www/website/images/example.png.
For local resources this is the prefered way of referencing them. This is a relative URL based on the document root (/var/www/mywebsite) of your website. This means when you have <img src="/images/example.png"> it will always resolve to /var/www/mywebsite/images/example.png.
If at some point you decide to switch domain it will still work because it is relative.
This is also a relative URL although a bit different than the previous one. This URL is relative to the current path. What this means is that it will resolve to different paths depending on where you are in the site.
For example when you are on the page http://yourdomain.com and you use <img src="images/example.png"> it would resolve on the server to /var/www/mywebsite/images/example.png as expected, however when your are on the page http://yourdomain.com/some/path and you use the exact same image tag it suddenly will resolve to /var/www/mywebsite/some/path/images/example.png.
When requesting external resources you most likely want to use an URL relative to the scheme (unless you want to force a different scheme) and when dealing with local resources you want to use relative URLs based on the document root.
An example document:
<link href='//fonts.googleapis.com/css?family=Lato:300italic,700italic,300,700' rel='stylesheet' type='text/css'>
<link href="/style/style.css" rel="stylesheet" type="text/css" media="screen"></style>
<img src="/images/some/localimage.png" alt="">
<script src="//ajax.googleapis.com/ajax/libs/jquery/1.10.2/jquery.min.js" ></script>
Does using absolute URLs load the page faster, than compared to using relative URLs? (Any time spent on resolving the relative path?)
Any difference possible would be so little it is not something you should worry about if it is measurable at all.
An example of including Google Jquery as protocolless: <script src="//ajax.googleapis.com/ajax/libs/jquery/1.10.2/jquery.min.js" ></script>
This answer assumes that absolute URLs aren't being generated dynamically which solves each issue mentioned.
J.Money is correct. Modern web frameworks have the notion of "reverse routing" to allow you to create URLS from one of your pages to another one of your pages (it has to be to another page in your same website). These allow you to give a name to a URL and then use this name instead of the URL. That way, if you ever want to change the URL, you can change the URL in one place, since everywhere else you only ever refer to that URL by its name.
You say "when you want to request a resource from an external domain over http you should use [http:// style]", then in the next paragraph you say protocol-relative links should "almost always be used when including external resources". Both paragraphs use images as examples. What's the difference?
what about ../path/to/image.png? Is this root-relative?
Yes that is relative to the current path @ReeceDodds. Say you are on https://foobar.com/my/awesome/subdir that URI points to https://foobar.com/my/awesome/path//to/image.png. It goes one level up from /my/awesome/subdir (..) and into /path/to/image.png.
Great answer, thanks a lot! For the sake of clarity, shouldn't you replace //yourdomain.com/images/example.png by //externaldomain.com/images/example.png ?
See this: http://en.wikipedia.org/wiki/URI_scheme#Generic_syntax
\ / \________________/\_________/ \__/ \___/ \_/ \_________/ \_________/ \__/
| | | | | | | | |
| userinfo hostname port | | parameter query fragment
| \_______________________________/ \_____________|____|____________/
scheme | | | |
| authority |path|
| | |
| path interpretable as filename
| ___________|____________ |
/ \ / \ |
urn:example:animal:ferret:nose interpretable as extension
An absolute url includes the parts before the "path" part - in other words, it includes the scheme (the http in http://foo/bar/baz) and the hostname (the foo in http://foo/bar/baz) (and optionally port, userinfo and port).
Relative urls start with a path.
Absolute urls are, well, absolute: the location of the resource can be resolved looking only at the url itself. A relative url is in a sense incomplete: to resolve it, you need the scheme and hostname, and these are typically taken from the current context. For example, in a web page at
you could put a link like so
<a href="pages/page1">click me</a>
In the href attribute of the link, a relative url s used, and if it is clicked, it has to be resolved in order to follow it. In this case, the current context is
so the schema, hostname, and leading path of these are taken and prepended to pages/page1, yielding
If the link would have been:
<a href="/pages/page1">click me</a>
(note the / appearing at the start of the url) then it would have been resolved as
because the leading / indicates the root of the host.
In a webapplication, I would advise to use relative urls for all resources that belong to your app. That way, if you change the location of the pages, everything will continue to work. Any external resources (could be pages completely outside your application, but also static content that you deliver through a content delivery network) should always be pointed to using absolute urls: if you don't there simply is no way to locate them, because they reside on a different server.
Relative URLs do not need to start with the URL path. //example.com/…, ?foobar and #foobar are also relative URLs and do not start with the URL path (well ok, for ?foobar you can say it does start with an empty path).
@Gumbo, are //example.com/…-type URLs called relative? that’s new to me.
@törzsmókus In terms of RFC 2396: “Relative URI references are distinguished from absolute URI in that they do not begin with a scheme name.”
There are really three types that should be discussed explicitly. In practice though URLs have been abstracted to be handled at a lower level and I would go as far as to say that developers could go through their entire lives without writing a single URL by hand.
Absolute URLs tie your code to the protocol and domain. This can be overcome with dynamic URLs.
Control - The subdomain and protocol can be controlled. People that enter through an obscure subdomain will be funneled into the proper subdomain. You can hop back and forth between secure and non-secure as appropriate.
Configurable - Developers love things to be absolute. You can design neat algorithms when using absolute URLs. URLs can be made configurable so that a URL can be updated site-wide with a single change in a single configuration file.
Clairvoyance - You can search for the people scraping your site or maybe pick up some extra external links.
Root Relative URLs tie your code to the base url. This can be overcome with dynamic URLs and/or base tags.
Root Relative Pros:
Relative URLs tie your code to the directory structure. There is no way to overcome this. Relative URLs are only useful in file systems for traversing directories or as a shortcut for a menial task.
<link src=“../.././../css/default.css” />
CONFUSING - How many dots is that? how many folders is that? Where is the file? Why isn't it working?
MAINTENANCE - If a file is accidentally moved resources quit loading, links send the user to the wrong pages, form data might be sent to the incorrect page. If a file NEEDS to be moved all the resources that are going to quit loading and all the links that are going to be incorrect need to be updated.
DOES NOT SCALE - When webpages become more complex and views start getting reused across multiple pages the relative links will be relative to the file that they were included into. If you have a navigation snippet of HTML that is going to be on every page then relative will be relative to a lot of different places. The first thing people realize when they start creating a template is that they need a way to manage the URLs.
COMPUTED - They are implemented by your browser (hopefully according to RFC). See chapter 5 in RFC3986.
OOPS! - Errors or typos can result in spider traps.
Developers have stopped writing URLs in the sense being discussed here. All requests are for a website's index file and contain a query string, aka a route. The route can be thought of as a mini URL that tells your application the content to be generated.
<a href="<?=Route::url('named_url', array('first' => 'my', 'last' => 'whacky'))?>">
Most people will make use of all three forms in their projects in some way or another. The key is to understand them and to choose the one best suited for the task.
what language / framework is used in the Routes example?
You're missing protocol-relative URLs, which are strictly better than completely absolute URLs. Absolute URLs are troublesome when upgrading the scheme (to HTTPS, mostly), relative URLs fix this.
@Tobu Just serve everything over HTTPS.
Side note: it looks like you used typographical quotes in most of your codes examples above. You might want to fix that.
And what type of url is the one with the dot first? E.g. "./index.html"
Could you elaborate a bit on Clairvoyance? If you were using "Root Relative" URLs, why wouldn't you be able to see people scraping your site?
@Lovethenakedgun If people copy code or even point their domain at your server's IP, relative URLs will keep users on the fake site while absolute URLs will lead users back to the real site. Absolute URLs are an attribution to the source. Search engines can be used to find careless scrapers or fraudsters that are stealing content or duplicating entire sites, such as for phishing.
For every system that support relative URI resolution, both relative and absolute URIs do serve the same goal: referencing. And they can be used interchangeably. So you could decide in each case differently. Technically, they provide the same referencing.
To be precise, with each relative URI there already is an absolute URI. And that's the base-URI that relative URI is resolved against. So a relative URI is actually a feature on top of absolute URIs.
And that's also why with relative URIs you can do more as with an absolute URI alone - this is especially important for static websites which otherwise couldn't be as flexible to maintain as compared to absolute URIs.
These positive effects of relative URI resolution can be exploited for dynamic web-application development as well. The inflexibility absolute URIs do introduce are also easier to cope up with, in a dynamic environment, so for some developers that are unsure about URI resolution and how to properly implement and manage it (not that it's always easy) do often opt into using absolute URIs in a dynamic part of a website as they can introduce other dynamic features (e.g. configuration variable containing the URI prefix) so to work around the inflexibility.
So what is the benefit then in using absolute URIs? Technically there ain't, but one I'd say: Relative URIs are more complex because they need to be resolved against the so called absolute base-URI. Even the resolution is strictly define since years, you might run over a client that has a mistake in URI resolution. As absolute URIs do not need any resolution, using absolute URIs have no risk to run into faulty client behaviour with relative URI resolution. So how high is that risk actually? Well, it's very rare. I only know about one Internet browser that had an issue with relative URI resolution. And that was not generally but only in a very (obscure) case.
Next to the HTTP client (browser) it's perhaps more complex for an author of hypertext documents or code as well. Here the absolute URI has the benefit that it is easier to test, as you can just enter it as-is into your browsers address bar. However, if it's not just your one-hour job, it's most often of more benefit to you to actually understand absolute and relative URI handling so that you can actually exploit the benefits of relative linking.
I'm going to have to disagree with the majority here.
I think the relative URL scheme is "fine" when you want to quickly get something up and running and not think outside the box, particularly if your project is small with few developers (or just yourself).
However, once you start working on big, fatty systems where you switch domains and protocols all the time, I believe that a more elegant approach is in order.
When you compare absolute and relative URLs in essence, Absolute wins. Why? Because it won't ever break. Ever. An absolute URL is exactly what it says it is. The catch is when you have to MAINTAIN your absolute URLs.
The weak approach to absolute URL linking is actually hard coding the entire URL. Not a great idea, and probably the culprit of why people see them as dangerous/evil/annoying to maintain. A better approach is to write yourself an easy to use URL generator. These are easy to write, and can be incredibly powerful- automatically detecting your protocol, easy to config (literally set the url once for the whole app), etc, and it injects your domain all by itself. The nice thing about that: You go on coding using relative URLs, and at run time the application inserts your URLs as full absolutes on the fly. Awesome.
Seeing as how practically all modern sites use some sort of dynamic back-end, it's in the best interest of said site to do it that way. Absolute URLs do more than just make you certain of where they point to- they also can improve SEO performance.
I might add that the argument that absolute URLs is somehow going to change the load time of the page is a myth. If your domain weighs more than a few bytes and you're on a dialup modem in the 1980s, sure. But that's just not the case anymore. https://stackoverflow.com/ is 25 bytes, whereas the "topbar-sprite.png" file that they use for the nav area of the site weighs in at 9+ kb. That means that the additional URL data is .2% of the loaded data in comparison to the sprite file, and that file is not even considered a big performance hit.
That big, unoptimized, full-page background image is much more likely to slow your load times.
An interesting post about why relative URLs shouldn't be used is here:
An issue that can arise with relatives, for instance, is that sometimes server mappings (mind you on big, messed up projects) don't line up with file names and the developer may make an assumption about a relative URL that just isn't true. I just saw that today on a project that I'm on and it brought an entire page down.
Or perhaps a developer forgot to switch a pointer and all of a sudden google indexed your entire test environment. Whoops- duplicate content (bad for SEO!).
Absolutes can be dangerous, but when used properly and in a way that can't break your build they are proven to be more reliable. Look at the article above which gives a bunch of reasons why the Wordpress url generator is super awesome.
When you say absolute URL do you mean the full url or using / to link to the base path? ie /products/wallets/thing.html as opposed to thing.html as opposed to http://www.myshop.com/products/wallets/thing.html
Prepending with a "/" will always be relative to the domain root, I believe. So if your domain is "www.example.com", any references coded as "/image1.jpg" would be interpreted as "www.example.com/image1.jpg". Items without the leading slash are interpreted as relative to the request root. When I say "absolute URL" I mean the fully qualified url. I cringe to send MSDN links through the internet but this is actually a pretty good breakdown: msdn.microsoft.com/en-us/library/windows/desktop/…
This is the current best practice although many people haven't realized it yet. I love Routes, such as in Kohana where you can use echo Route::url('route_name') to build an absolute URL using the site URL and route information with the option to make it over HTTPS.
I feel the need to point out that dialup modems weren't left back in the 80's. There are plenty of people now who have no choices other than dialup or super overpriced very limited satellite internet... and if you're poor, what could be better than free dialup? It really bothers me to see developers that believe that dialup doesn't exist anymore. It can take 5-10 minutes(!!!) to login to my bank's website on dialup... Paypal, Amazon and Ebay aren't much better. Egg Cave (a virtual pet site) and Facebook just plain don't work on dialup. That affects a lot of people who live in rural areas.
Okay, yes, while there are some people on dialup, the vast (vast) majority of users are not on dialup. It also has a lot to do with demographic. If you're shooting for mega-huge ultra-performant results, then start clipping your domain off your URL. But the main point of that comment was to underscore that performance is usually found in other areas - you could gain all of the transfer time that you lost in url bytes by optimizing a single image. but, if 20% or more of your users are on dialup, I guess thats important. In 2015, and for all practical purposes, that's just not the case.
I might add, that ebay and most bank websites are not very well optimized. In fact, they suck entirely. But, they are also coming out of enormous frameworks run by hundreds or thousands of people, the management of which would be impossible without a considerable level of complexity. Not to justify the fact that the current version of the Microsoft.com homepage, which has merely a few blades on it, has 384kb of source code to it (that's just ridiculous), but it's not always quite so black-and-white, either. So yes, you're right, strive for performance, but keep best practices in mind, too.
Generated or not is orthogonal to the type of URL. You're arguing for generated, but user agents don't care whether the generated urls are absolute.
I'm not arguing for generated for any reason other than the fact that automation makes life easier. I'm certainly not saying that a UA would ever know the difference.
It seems to me that the linked article and most part of your answer say "relative URLs are bad because of SEO". To me, from a technical point of view, I don't see anything wrong with them, however. More than that, I don't see a merit in having to rely on a tool or CMS mechanism in order to be able to maintain my URLs -- especially not if mainly for the sake of SEO. Simplicity always wins for me, and that means relative URLs win over absolute URLs any day.
I would heartily recommend relative URLs for pointing bits of the same site to other bits of the same site.
Don't forget that a change to HTTPS - even if in the same site - is going to need an absolute URL.
You're wrong about HTTPS.
A URL that starts with the URL scheme and scheme specific part (http://, https://, ftp://, etc.) is an absolute URL.
Any other URL is a relative URL and needs a base URL the relative URL is resolved from (and thus depend on) that is the URL of the resource the reference is used in if not declared otherwise.
Take a look at RFC 2396 – Appendix C for examples of resolving relative URLs.
If it is for use within your website, it's better practice to use relative URL, like this if you need to move the website to another domain name or just debug locally, you can.
Take a look at what's stackoverflow is doing (ctrl+U in firefox):
<a href="/users/recent/90691"> // Link to an internal element
In some cases they use absolute urls :
<link rel="stylesheet" href="http://sstatic.net/so/all.css?v=5934">
... but this is only it's a best practice to improve speed. In your case, it doesn't look like you're doing anything like that so I wouldn't worry about it.
Let's say you have a site www.yourserver.com. In the root directory for web documents you have an images sub-directoy and in that you have myimage.jpg.
An absolute URL defines the exact location of the document, for example:
A relative URL defines the location relative to the current directory, for example, given you are in the root web directory your image is in:
(relative to that root directory)
You should always use relative URLS where possible. If you move the site to www.anotherserver.com you would have to update all the absolute URLs that were pointing at www.yourserver.com, relative ones will just keep working as is.
In most instances relative URLs are the way to go, they are portable by nature, which means if you wanted to lift your site and put it someone where else it would work instantly, reducing possibly hours of debugging.
There is a pretty decent article on absolute vs relative URLs, check it out.
In general, it is considered best-practice to use relative URLs, so that your website will not be bound to the base URL of where it is currently deployed. For example, it will be able to work on localhost, as well as on your public domain, without modifications.
+1 I agree. There might be (a few) times when absolute urls are better, for instance when using a CDN, or if you need to change the content website. Searching for a domain name is a lot easier than searching for relative urls IMHO.
For maintenance purposes it can be easier to use absolute URL's, without the domain name. i.e. On StackOverflow use the absolute URL '/questions/2005079/absolute-vs-relative-urls' to link to this question. The '/' on the front makes the URL absolute. This approach pays off when you go to move your files around or change the directory structure of your project.
The localhost argument is a good one, but when moving a site from one domain to another, a simple 'find & replace' action in TextMate, Coda or your favorite editor will solve it
can you provide any sources? I've only read ever that absolute url's are a better idea. Also, you need to have some kind of system to make absolute urls anyways if you have an RSS feed.
I know this answer is relatively old, but it still actual. If you can, please tell your opinion about this article, as it's opposite to you answer: yoast.com/relative-urls-issues
@Baumr But can you do that? I'm definitely a fan/proponent of using that method, but get into a bigger framework and large refactorings are a notoriously daunting/terrifying task. While you thought you may have switched them all, even in a smart IDE, often times they can be missed if they're coded into strings or dynamically created. The worst part of that is that usually those missed references aren't caught until after your solution goes back into production... :(
I had the case that I used absolute URL's started with / so moving to another server should be possible, but then I deployed my app in www.mydomain.com/apps/ and now i can't find any resources ofc. Thats why I agree that relative URL's are safer - still they can messup when you reuse serverside templates in different routes, but there you can easily use header-redirects to send the user to the correct path before rendering.
This is simple and better explanation and +1!
@Baumr 'find & replace' is no solution at all if your product is a deployed website which is shipped to customers
@Mike why would you call root-relative URLs ‘absolute’?
@törzsmókus good question. It doesn't let me edit and that that was years ago, before I even came across the term root relative.
To use absolute links may try <a href="<?php echo $_SERVER['REQUEST_SCHEME'];?>://<?php echo $_SERVER['SERVER_NAME'];?>rest-of-url"
<a href="<?php echo $_SERVER['REQUEST_SCHEME'];?>://<?php echo $_SERVER['SERVER_NAME'];?>rest-of-url"
This is the absolute best answer to the question, relative to the other answers :) Furthermore I would point out that a number of SEO experts warn that absolute URLs are somehow better and that digital marketing executives should argue with developers and ask them to switch all URLs from relative to absolute. This one factor convinces me that relative URLs are in fact the way to go.
Can a URL even be relative? Isn't the presence of the protocol what makes a URI and URL?
This isn't even answering the question of what the difference is...
an old but revelating answer on the terminology: stackoverflow.com/a/904066/501765 (relative path, absolute path, relative URL, absolute URL)