[icfp] size of elab fucntions

icfp-2016/fig-stats.tex

@@ -1,15 +1,52 @@
 \newcommand{\twoline}[2]{\parbox[s]{1.44cm}{\flushright\hfill #1\newline#2}}
 \begin{tabular}{l r r r}
-  Module             & LOC & $\interp$ & $\trans$   \\\hline
+  Module             & LOC & $\interp$ (LOC) & $\elab$ (LOC)  \\\hline
   \mod{db}           & 263 &   2  (78) &  2  (101)  \\
   \mod{format}       &  66 &   1  (33) &  1 \,~(21) \\
-  \mod{function}     &  31 &   1 ~~(8) &  1 \,~(11) \\
+  \mod{function}     & 117 &   1  (11) &  2 \,~(51) \\
   \mod{math}         &  90 &   1 ~~(3) &  5 \,~(46) \\
-  \mod{regexp}       & 122 &   6  (60) &  5 \,~(33) \\
+  \mod{regexp}       & 137 &   6  (60) &  5 \,~(33) \\
   \mod{vector}       & 228 &   1  (19) & 13  (163)  \\\hline
-  {\bf Total}        & 800 &  12 (201) & 27  (375)  \\
+  {\bf Total}        & 901 &  12 (204) & 27  (415)  \\
 \end{tabular}
 %% AVG
 % loc : 228.5
 % int : 33.5
 % trn : 62.5
+
+%% elab frontends
+%% mean 10, median 7
+%db
+%- query 35
+%- connect 7
+%
+%fromat
+%- fmt 16
+%- pf 5
+%
+%curry
+%- c 12
+%- m 40
+%
+%math
+%- 1 1 1 1
+%- e 8
+%
+%rx
+%- 1 1 1 1
+%- m 16
+%
+%v
+%- len 5
+%- ref 10
+%- set 10
+%- map 19
+%- map! 15
+%- app 29
+%- ls 14
+%- im 8
+%- fll 13
+%- t 2
+%- tr 2
+%- d 2
+%- dr 2

icfp-2016/implementation.scrbl

@@ -11,7 +11,7 @@
 @; TODO this is all so boring right now, need to revise heavily
 
 @Figure-ref{fig:stats} gives a few statistics regarding our implementation.
-The purpose of this section is to explain why the numbers are low.
+The purpose of this section is to explain why the line counts are low.
 @; Ode to macros, implementation a symphony
 
 @figure["fig:stats"
@@ -19,16 +19,16 @@ The purpose of this section is to explain why the numbers are low.
   @exact|{\input{fig-stats}}|
 ]
 
-In total, our six applications comprise 800 lines of code (LOC).
+In total, the code for our six applications described in @Secref{sec:usage}
+ comprise 901 lines of code (LOC).
 Another 145 lines implement common functionality, putting the grand total
- just under 1000 LOC.
+ just over 1000 LOC.
 
 Except for @exact|{\mod{db}}| and @exact|{\mod{regexp}}|, each of the
- core modules defines a single function in @exact|{$\interp$}|.
+ core modules defines a single interpretation function (in @exact|{$\interp$}|).
-In @exact|{\mod{db}}| the two functions are the schema predicate and @tt{SQL}
+In @exact|{\mod{db}}|, the two functions are the schema predicate and @tt{SQL}
- query parser (we omit the trivial interpreter for connections).
+ query parser.
-    @; TODO no parentheses?
+In @exact|{\mod{regexp}}|, we have six group-parsing functions
-On the other hand, @exact|{\mod{regexp}}| implements six group-parsing functions
  to match the six string-like input types
    @;@note{Strings, Regex literals, Posix Regex literls, and byte-string variations of each.}
  accepted by Racket's @racket[regexp-match].
@@ -39,11 +39,13 @@ The smallest interpreter is the composition of Racket's @racket[number?] predica
 The largest is the query parser (35 LOC), though the analyses for
  format strings and regular expressions are approximately the same size.
 
-The @exact{$\trans$} functions are aliases for standard library procedures.
+The elaboration functions are aliases for standard library procedures.
-   @; TODO much better to show off short code. But let's draft it first.
+Typically, these functions match a syntactic pattern, check for value errors,
-In many cases we are able to re-use code between similar functions.
+ and elaborate to a specialized Typed Racket procedure call.
-For instance, the arithmetic operators @racket[+ - * /] are implemented by
+All these tasks can be expressed concisely; the average size of a function in
- a single fold.
+ @exact{$\elab$} is 10 lines and the median is 7 lines.
+Much of the brevity is due to amortizing helper functions, so we include helpers'
+ line counts in the figure.
 
 
 @; -----------------------------------------------------------------------------