ABSTRACT
Web based applications have gained increased popularity over the
past decade due to the ubiquity of the web browser platform across
different systems. Rapid evolution of web technologies has lead
to inconsistencies in web browser implementations, which has resulted
in cross-browser incompatibilities (XBIs) – situations in which
the behavior of the same web application varies across browsers.
Such XBIs can either result in cosmetic defects or completely prevent
the user from accessing part of the web application’s functionality.
Although XBIs are a fairly common problem faced by
users of web applications, there hardly exist any tools for discovering
them automatically and assisting the fix of such problems. Due
to this, developers have to invest considerable manual effort in detecting
XBIs, which is both time consuming and error prone. This
paper presents our technique and tool – CROSSCHECK to automatically
detect and report XBIs to the web developer along with debug
information about the affected HTML element, thereby helping the
developer to fix the issue. CROSSCHECK is the first technique to
practically discover XBIs in real web applications and in doing so,
it combines concepts from graph theory, computer vision and machine
learning. Our results show that CROSSCHECK 1
is practical
and can effectively find both cosmetic and functional XBIs in real
world web applications.