summaryrefslogtreecommitdiff
path: root/paper/disc.tex
diff options
context:
space:
mode:
Diffstat (limited to 'paper/disc.tex')
-rw-r--r--paper/disc.tex16
1 files changed, 16 insertions, 0 deletions
diff --git a/paper/disc.tex b/paper/disc.tex
new file mode 100644
index 0000000..0fdee9f
--- /dev/null
+++ b/paper/disc.tex
@@ -0,0 +1,16 @@
+\section{Future work}
+\label{sec:disc}
+
+As we have seen in \autoref{sec:interp:cond}, not all rules that can be
+specified in mathematical notation are trivial to translate to a functional
+program. It should be investigated under what conditions rules can or cannot be
+translated directly. What kind of functional tools can we come up with to make
+this easier?
+
+Our claim has been that semantic rules can be translated almost directly to an
+implementation in a functional language. We can then ask ourselves: is it
+possible to store rules in a data structure and write a universal interpreter,
+say \CI{run :: [Rule stm state] stm state -> Either Error state} (note the parametrization of
+\CI{stm} and \CI{state})? Under what conditions is it possible for such a
+universal interpreter to choose those rules that allow for the most efficient
+interpretation?