The work of Gordon S. Novak at the University of Texas at Austin
on specializing generic algorithms is related to our work
on adaptive software. He writes in

@ARTICLE{novak:generic-algs,
  AUTHOR = "Gordon S. Novak",
  TITLE = "Creation of Views for Reuse of Software with Different Data
  Representations",
  JOURNAL = ieee-se,
  YEAR = 1995,
  PAGES = "993-1005",
  MONTH = "Dec.",
  VOLUME = 21,
  NUMBER = 12
  }

\begin{quote}
Traditional data types combine two issues that should be separated
to facilitate reuse 1) the way in which application data are represented
2) procedures associated with a conceptual kind of object. ...
When the two are combined, the code of a procedure implicitly states
assumptions about details of data representation. This inhibits
software reuse ... Object-oiented-programming and functional programming
partly separate the two issues ...
\end{quote}

Novak developed the GLISP 
compiler which can specialize a generic procedure
by compiling it relative to a view. Novak's work on structure-shy
programming is not based on 
succinct subgraph specifications but uses other techniques,
such as equation solving.