tags:

views:

149

answers:

2

Hello. Say one's writing a "browser" and wants to test it on many pages with broken encoding, invalid HTML, missing meta tags or headers etc. An obvious solution is to get a big pile of URLs and test the application on each of them, which takes a lot of time and doesn't guarantee that the first 1000 urls were ok, while the next 100 will be broken. So is there a site providing many sorts of different broken pages for testing?

update:

sorry, the original "web application" was misleading, I meant a desktop browser-like application

update 2:

one of the functions is filling HTML forms so I'd like to test, whether the url-encoding was correct, whether all the fields got filled in and so on

update 3:

another example may be, there is a submit button with foreign value encoded in Shift-JIS and if my app renders it bogus and sends wrong POST request string (which is what ruby mechanize still does), the site notifies me that a standard-compliant app would encode the string in another way

+2  A: 

You can try the acid tests, they are to ensure your program is standards compliant. It's not exactly what you asked for but it's a step in the right direction. Most browsers handle broken pages and quirks in different ways and people don't often want to test to see if their broken site renders brokenly in different browsers so I don't think there is going to be something specifically for you but it wouldn't be that hard to do yourself if you wanted to specifically break your program.

Eg.

<sCriPt
type="text/javascript"
>
/*    <span class="*/>awesome"> // */
</script>
<b <i>>a</></b>
<body type="muscular"></body>
<! text
--

I'd like to point you to Adobe Air, I think it would serve your purposes:

Adobe AIR is a cross-operating system runtime that enables you to use your existing HTML/Ajax, Flex, or Flash web development skills and tools to build and deploy rich Internet applications to the desktop.

Sam152
+1  A: 

Assuming that you're building a web-browser, I'd suggest using your own computer as a server, using Apache, Lighttpd (whether Windows, Linux or Mac); that would be the fastest way, surely?

To create a list of 'broken' pages, use the W3 Validator to validate various pages online, view-source and then save as html, removing various closing tags, nesting <p> tags, removing var-names or semi-colons from JS scripts, using the html 4.1 under an xhtml-strict doctype, etc.

The benefit of this is that you know how they're broken in advance, and what the expected outcome should (or is likely to) be. Also, you don't have to worry (too much) about the network's effect on the render-time.

David Thomas