I'm not supporting clients that can't run JavaScript.
I want my URL paths to look like /settings, not /#settings, i.e., no hash or fragment identifiers.
I understand how to intercept a click event with JavaScript.
But, what should the server do? Should the server just reply to every page, e.g., /, /settings, /profile, etc., with the same exact HTML file?
Then, after the page loads, the JavaScript will decide which parts of the HTML to display based on the location's path?
You need to use pushState. Here's a link to MDN on how to do it.
https://developer.mozilla.org/en-US/docs/Web/API/History_API#The_pushState()_method
Related
What actually happens when you enable the html5 mode ? This might go back to the point how routing takes place in single page applications
What I perceived before(it may be wrong) : Looking at the dirty url in the angularjs application I always assumed it being url fragment to which different views are bind for different fragments. So in short we already have all the pages and a particular fragment is being displayed for a particular url .
Now in order to remove the hash you have to set html5mode true and you have to tell the server to return the index page for every request other than your apis . Kinda like
app.get('/ap1',some);
//more apis
*
*
*
//in the end
app.get('*' ,(req,res,next) => req.sendFile('index.html'));
Shouldn't the request go the server and the page be reloaded everytime the url changes ?
and what does html5mode does to the browser ? In the newer frameworks like react and angular(2 or greater) , you don't even have to enable html5mode(except in angular 2 where you have to tell what kind of url you want) .
What is this sorcery ?
This uses the History API.
It is designed so that developers can:
Tell the browser things like "I have used JavaScript to change the contents of the page. It is now the same as the page you would get if you asked for /other/page"
Intercept clicks on Back and Forward so that they can use JS to change the contents of the page to the be the same as the URLs that would be navigated to if the click on Back/Forward wasn't intercepted.
This updates the browser history and the URL in the address bar without triggering a normal navigation to a fresh page. The idea is that you can provide a smoother, faster transition between pages while still having real URLs that work well if you link to them or come back to them later.
That way:
If the JavaScript fails for any reason, the real URLs can still deliver the page
If a search engine indexes the URL, it gets the real data from it
If a visitor lands on a URL that isn't the homepage, they get the content they asked for immediately without having to wait for JavaScript to run, making additional Ajax requests, and assemble all the pieces client side.
It was designed in response to people using hangbangs instead of real URLs.
Now in order to remove the hash you have to set html5mode true and you have to tell the server to return the index page for every request other than your apis
This is terrible advice from Angular.
It abandons all of the good stuff that the history API can provide in order to have all the drawbacks of hashbangs, but with nicer looking URLs.
The problem with doing it properly is that it requires duplicating the logic on the server and the client, which is a lot of work. Isomorphic JavaScript is an approach to reduce this workload.
and what does html5mode does to the browser ?
Nothing. The html5mode variable is read by Angular. The browser doesn't care about it directly. It tells Angular to use the History API instead of hangbang URLs.
In the newer frameworks like react and angular(2 or greater) , you don't even have to enable html5mode(except in angular 2 where you have to tell what kind of url you want) .
They just use the History API by default.
What actually happens when you enable the html5 mode ? This might go back to the point how routing takes place in single page applications
What I perceived before(it may be wrong) : Looking at the dirty url in the angularjs application I always assumed it being url fragment to which different views are bind for different fragments. So in short we already have all the pages and a particular fragment is being displayed for a particular url .
Now in order to remove the hash you have to set html5mode true and you have to tell the server to return the index page for every request other than your apis . Kinda like
app.get('/ap1',some);
//more apis
*
*
*
//in the end
app.get('*' ,(req,res,next) => req.sendFile('index.html'));
Shouldn't the request go the server and the page be reloaded everytime the url changes ?
and what does html5mode does to the browser ? In the newer frameworks like react and angular(2 or greater) , you don't even have to enable html5mode(except in angular 2 where you have to tell what kind of url you want) .
What is this sorcery ?
This uses the History API.
It is designed so that developers can:
Tell the browser things like "I have used JavaScript to change the contents of the page. It is now the same as the page you would get if you asked for /other/page"
Intercept clicks on Back and Forward so that they can use JS to change the contents of the page to the be the same as the URLs that would be navigated to if the click on Back/Forward wasn't intercepted.
This updates the browser history and the URL in the address bar without triggering a normal navigation to a fresh page. The idea is that you can provide a smoother, faster transition between pages while still having real URLs that work well if you link to them or come back to them later.
That way:
If the JavaScript fails for any reason, the real URLs can still deliver the page
If a search engine indexes the URL, it gets the real data from it
If a visitor lands on a URL that isn't the homepage, they get the content they asked for immediately without having to wait for JavaScript to run, making additional Ajax requests, and assemble all the pieces client side.
It was designed in response to people using hangbangs instead of real URLs.
Now in order to remove the hash you have to set html5mode true and you have to tell the server to return the index page for every request other than your apis
This is terrible advice from Angular.
It abandons all of the good stuff that the history API can provide in order to have all the drawbacks of hashbangs, but with nicer looking URLs.
The problem with doing it properly is that it requires duplicating the logic on the server and the client, which is a lot of work. Isomorphic JavaScript is an approach to reduce this workload.
and what does html5mode does to the browser ?
Nothing. The html5mode variable is read by Angular. The browser doesn't care about it directly. It tells Angular to use the History API instead of hangbang URLs.
In the newer frameworks like react and angular(2 or greater) , you don't even have to enable html5mode(except in angular 2 where you have to tell what kind of url you want) .
They just use the History API by default.
I have a client side router with which I can navigate through my web application with hash based urls, like http://example.com/#home. With the history.pushState(..) method I make the url more cleaner, like http://example.com/home. Now my problem is when someone reload the page, the url is unknown, because there is no hash in it. Is there a elegant way to fix this?
What it's meant for
history.pushState(..) is needed for single-page-applications to create an artificial browser history. That artificial history allows users to navigate via back and forward buttons and also to bookmark a "page" (there is only one, but it feels like multiple pages).
What you're doing
Now you're rewriting URLs to locations, which don't exist. On reload the browser navigates to that location and finds nothing, it doesn't know the relationship between your single-page-application and that new URL.
Solutions
You could try to fix this by creating redirections on server-side. Then you'll be able to read the current state. There are multiple options for redirects:
mod_rewrite - all to index except static folder
Redirect from an HTML page? (discouraged)
...
Keep the hashed URLs and use the saved time to work on real features.
I'd take the second option.
I'm looking for a way to support SEO for content which is loading via AJAX. My site handles urls for example: www.example.com/part/{number}. Then on a homepage JS executes AJAX request and receives a content. Inside the AJAX callback the pushState is used.
Will the Googlebot respect dynamic content (without duplicate) if the homepage has always the same content? If not, are there ways to do that?
Will the Googlebot respect dynamic content (without duplicate) if the homepage has always the same content?
Probably not (although they do pay some attention to JS these days).
If not, are there ways to do that?
Use pushState properly. When you change the URL with it, have the JS change the DOM to reflect what the server would deliver if the URL was requested.
Don't load the default content, then examine the URL with client side code, then modify it. That defeats the object of pushState and you might as well go back to the bad old days of hashbangs.
You may have seen app.asana.com.
If not you should check it out, it is a very nice designed webapp.
But I can't figure out how they handle the whole URL management.
Backbone.js or Knockout.js handles the URL with the #, and everything after that is just generated.
But asana doesn't have a hash and can modify the URL, how are they doing this?
Looks like they're using HTML5 history.pushState(); so they don't have to refresh the page and so they don't have to use # (hashes) in the URL to go to a certain part in a web app.
Here's a good tutorial about history.pushState();: https://developer.mozilla.org/en/DOM/Manipulating_the_browser_history
This is the what Google+ and Facebook uses to change the URL without refreshing.
I hope this helps.
HTML5 Push State: http://spoiledmilk.dk/blog/html5-changing-the-browser-url-without-refreshing-page
The big benefit here is that if you paste an Asana URL directly into the browser (or click on a link from an email), the server sees the full URL and can immediately send the appropriate task data to the client. We used to use url fragments, but we needed to do a second round trip after the application loaded to read the fragment in JavaScript and pass it to the server.