Difference between revisions of "AppSecEU08 Input validation: the Good, the Bad and the Ugly"

From OWASP
Jump to: navigation, search
 
(2 intermediate revisions by one user not shown)
Line 5: Line 5:
 
The discussion is illustrated by a case study
 
The discussion is illustrated by a case study
 
of a well-intentioned but flawed attempt at implementing meticulous
 
of a well-intentioned but flawed attempt at implementing meticulous
input validation. Architectural
+
input validation. The application's validation code is scattered throughout the code base.
refactoring, arguably mundane, but often neglected, will be employed to get rid of the
+
I investigate what can be done to improve the code.
worst code smells. For those who favor life on the bleeding edge, the
+
However, writing elegant validation code is found to be very hard in current mainstream technologies, so I also explore some promising alternatives.
discussion also ambles into some more tentative territory.
+

Latest revision as of 14:41, 22 April 2008

This talk discusses input validation design choices and recommends practices that provide developers a fighting chance to survive architectural decay as an application matures.

The OWASP 2004 Top Ten adviced never to trust user input. Although fundamentally sound, it led to many maintenance nightmares and insecure web applications. This talk argues that the enthusiasm for input validation must be tempered by a resolve to eliminate code duplication to maintain sanity and security. I will show this is possible, even in the face of apparently conflicting objectives, namely usability and protection against malicious users.

The discussion is illustrated by a case study of a well-intentioned but flawed attempt at implementing meticulous input validation. The application's validation code is scattered throughout the code base. I investigate what can be done to improve the code. However, writing elegant validation code is found to be very hard in current mainstream technologies, so I also explore some promising alternatives.