The cornerstone of every software development environment is the source code---the most important component of the machine-to-human interface. It is structured according to rules abstracted from natural languages and this feature is the key to the success of the human interface based on it. Due to the increasing complexity of the software systems, new requirements on the source code have arisen because there is an increasing need to automate its processing in various ways. Therefore it is important for it to allow the processing tasks to be easily formalized and so made ready for automatic execution. Also, it has to allow easy addition and processing of arbitrary metadata needed by the development process. The new requirements can be met by adopting a new textual format that complies with the XML standard. The main problem in the transition is the design of the new source code syntax and it was found that in the community dealing with this problem there is a lack of a sufficiently systematic approach and that the key concept of the abstract syntax is being misinterpreted. Therefore a systematic process was conducted which identified the exact requirements on the syntax and the principles that need to be followed in its design to meet those requirements. A syntax for the Java programming language was developed and compared to two existing syntaxes. It was shown that the proposed syntax meets the requirements to a significantly higher proportion. The source of the deficiencies of the alternative syntaxes was identified as lying in their divergence from the mentioned design principles, which confirms their validity.