[icfp] experience report

icfp-2016/experience.scrbl

@@ -1,18 +1,34 @@
 #lang scribble/sigplan
 @require["common.rkt"]
 
+@; TODO experiment is BOGUS because it doesn't measure overhead of define/let
+@; Make those automatic and try again!
+
 @title[#:tag "sec:experience"]{Experience}
-@; compile times, running times, binary size
-@; LOC reduction, bugs caught
 
-Coarse measure of usefulness, applied to xxx lines of code.
+Our initial experience programming with the library has been positive.
+By far the most useful application is to @racket[regexp-match], as
+ Typed Racket's default requires either a type cast or guards on each
+ matched group.
+The bugs reported by @racket[printf] and others are also useful, though
+ the real-time difference between catching a @racket[printf] bug during compilation
+ versus finding the same bug at run-time is small, maybe 1-3 seconds.
+We have also observed performance improvements on microbenchmarks involving
+ maps over large vectors, though we have yet to find the same benefits in real
+ code.
 
-We used the library to simplify @todo{lines} of database code and @todo{lines}
- of vector code.
-Performance of the vector code improved by @todo{xxx}, a nice bonus.
+In total, we have applied the library to 10,000 lines of Typed Racket code
+ taken from 7 small projects.
+Including the library is a 1-line change, but the programmer needs to remove
+ now-redundant type annotations and casts manually.
+Compiling with our library adds no statistically significant overhead, but
+ tends to produce slightly larger bytecode files (at most 2% larger).
+Running times are slightly worse for the added casts; in the worst case
+ we saw 2-second slowdown.
+This could be improved by tighter integration with the type checker.
 
-@todo{figure} gives microbenchmarking results for printf, regexp, and vector-@todo{}.
-Have compile-time, executable size, and run-time.
-Things look okay.
-We don't bother trying to quantify the bug-catching utility.
 
+@;@figure["fig:experience"
+@;  "Experience Report"
+@;  @exact|{\input{fig-experience}}|
+@;]

icfp-2016/fig-stxclass.tex

@@ -1,6 +1,6 @@
 \begin{center}
 \begin{tabular}{l l l}
-  \emph{Syntax Class}  & \emph{Purpose} \\\hline
+  Syntax Class  & Purpose \\\hline
   \mod{fun/arity}      & Infer function arity \\
   \mod{num/value}      & Evaluate a numeric expression \\
   \mod{pattern/groups} & Count regexp groups \\

icfp-2016/intro.scrbl

@@ -7,6 +7,7 @@
 @; - a note on macro-land? where truthy & boolean monads are king?
 @;   - also, the untyped style of reasoning, the needing to unfold code
 @; - transf vs. macro
+@; - remove "OUR"
 
 
 @require["common.rkt"]

icfp-2016/usage.scrbl

@@ -253,6 +253,8 @@ Our implementation employs the first strategy, but generates a new @racket[curry
  at each call site by folding over the inferred function domain.
 @; Mention this later, or show code?
 
+TODO same goes for zipWith in a language without polydots
+
 
 @; =============================================================================
 @section{Constant Folding}