User responsibilities are basically whether or not to trust your content to run in their machines, leading me back to the security you, the developer, will implement. This ability to trust your content and run or not run your scripts is within the capabilities of the browser; it is a setting in the preferences section of the browser.
The Problem with Frames
When the first instances of frames were implemented on the Internet, it took all of two months for the first major security holes to open up and be recognized by the hacking community. To understand how frames were used to present a security hole is to first understand frames themselves.
The way a web page renders frames is as follows:
- A page with instructions for the frames to be rendered is loaded.
- The addresses and particulars of how the frames will be rendered such as border properties and the placement of the frames themselves is given to the browser.
- The documents that will be used to display the content of the frames are then loaded into the given frames.
- The page is then rendered (displayed) to the browser window.
The Same Origin Policy: Security the Internet Explorer Way
- image: The lowsrc and src properties.
- layer: The src property.
- location: Every property available to location except location.x and location.y.
- window: The find property.
Occasionally, you may have a need to go against the Same Origin Policy to achieve the desired result from your web application. An exception has been made to address this possibility. To access the information within a page created and displayed within one of the page frames, you would use the "document.domain" statement to list a domain that is trusted by the web application and thus the browser. For example, for a page that originated at http://developer.walkthegeek.com to access variables and scripted entities from a page that originated at http://www.walkthegeek.com, you would use the following statement within your function:
document.domain = "walkthegeek.com";
Setting the document.domain property in this way tells the browser to trust all content from the domain www.walkthegeek.com as well as all subdomains of walkthegeek.com, such as the aforementioned developer.walkthegeek.com.
Data Tainting: Security the Netscape Way
Because the Data Tainting is part of your operating system, it cannot be turned on and off through the browser. Examine the following list that states how to enable Data Tainting in the various operating systems. It's really quite easy to do.
- Windows - NS_ENABLE_TAINT=1: Place the statement given in the autoexec.bat file for Windows 3.1X, 95, 98, and NT. For Windows NT, you also may set data tainting as a User Environment variable.
- UNIX - NS_ENABLE_TAINT=1: This one depends on which UNIX shell you are operating with. Basically, you would set an environment variable through the use of the set env or env commands.
- Macintosh: You would remove the two forward slashes before the NS_ENABLE_TAINT statement, which can be found by editing the resource of type envi and number 128 in the Navigator application itself. The NS_ENABLE_TAINT statement should be near the end of the document.
- OS/2 - NS_ENABLE_TAINT=1: Set the given statement in the config.sys file in the root of your startup drive.
- document: The cookie, domain, lastModified, links, referrer, title, and URL properties
- form: Every instance of a form element
- history: The current, next, previous, and toString methods and properties
- link: The hash, host, hostname, href, pathname, port, protocol, search, and toString properties and methods
- location: The hash, host, hostname, href, pathname, port, protocol, search, and toString properties and methods
- option: The defaultSelected, selected, text, and value properties and methods
- plugin: The name property
- window: The defaultStatus, name, and status properties and methods
Netscape's Page Signer tool will allow you to create your individual and unique online security identification. The tool builds a JAR (Java Archive) file that includes both your security certificate and your code. When the document has an HTML SCRIPT tag that has the ARCHIVE attribute set, the browser performs a verification check before the code is executed. An alert box pops up and gives the user a chance to accept or decline the running of your script. If the script is included within the document and not within an external .js file, the JAR file should include only your security identification, although it is still accessed through the use of the ARCHIVE attribute.
Because the user can accept or decline a script from running, Netscape has provided another level to the security that can be enforced by the user. Some think it to be too user-driven to use often. By this I mean that the browser will ask too many questions for what areas of the script to be run, causing increased user annoyance. The areas of the script that are available to be accepted or declined through the use of an alert box are set using the Java method "netscape.security.PrivelidgeManager.enablePrivelidge()". Here is a list of options that a developer can attempt to have the user verify:
- UniversalBrowserAccess: This method allows both the reading and the writing of priviledged data in and to the browser.
- UniversalBrowserRead: This method allows only the reading of privileged data in the browser, and is required when using the history object or getting the value of a DragDrop value within the browser.
- UniversalBrowserWrite: This method allows only the writing of privileged data in the browser, and is required when using any property of an event object, adding or removing any of the browser's content bars (status bar, menu bar, and so forth), and setting the window object's values within the script.
- UniversalFileRead: This method allows the reading of the file system of the user's machine, and is required when using the fileUpload() method.
- UniversalPreferencesRead: This method allows the script to read and report the browser's preferences settings.
- UniversalPreferencesWrite: This method allows the script to set the preference settings within your (the user's) browser.
- UniversalSendMail: This method allows the script to send an email with the user's name, and is required when using the news: or mailto: attributes within a script.
You can see that the aforementioned pros and cons are quite evident here. Additionally, each of the two major browsers—Microsoft's Internet Explorer and Netscape's Navigator—have differing security measures in place. Internet Explorer uses the Same Origin Policy, whereas Netscape uses the Signed Scripts method. Each is good, in its own way. What is frustrating, as you'll see when you attempt to enforce security, is that they are very different. A totally different directory path for each is sometimes required, effectively doubling the work you'll have to put in to satisfy the security considerations for each browser.