Recent versions of Mozilla and Netscape include support for Hebrew, Arabic, and other Bidi languages. The implementation was contributed by an international team based in the IBM centers in Austin, Cairo and Jerusalem.

This paper discusses in detail the technical challenges that faced the development project and their solutions. In particular, it describes the challenge of developing an implementation that would be compliant with the Unicode, HTML, DOM, and CSS standards and produce correct results on all the different platforms supported by Mozilla, both with and without native Bidi support. Web content creates special problems for Bidi processing, partly because of the inherent complexity of mark-up text, and partly because of the existence of "Visual" pages designed for older browsers without Bidi support. Support was also required for the entry of Bidi text in input fields and the HTML composer, and for the possibility of a localized version for the Hebrew or Arabic market with right-to-left UI.

The actual implementation of the Unicode Bidi Algorithm was taken from IBM's open source International Components for Unicode, with some necessary modifications, and the initial design for incorporation of Bidi support into the existing architecture of Mozilla's parser, content model and frame model was evolved and refined in discussions with the online Mozilla community.

The history of this project is an interesting case study in itself, and may also serve as an example for future extensions to the international language support of Mozilla/Netscape.