checkpoint

ParsingIntroduction.lhs

@@ -17,28 +17,13 @@ favourite editor, and use ghci to experiment. To download, use:
 git clone https://github.com/JakeWheat/intro_to_parsing.git
 ```
 
-TODO: experiment: create wrapper modules for all the interesting stuff
-in parsec, which fixes the parser types to use
-Text.Parsec.String.Parser. This should make things much simpler for
-beginners, especially reading the code and dealing with error
-messages.
-
-modules:
-parse, ParseError from somewhere
-Text.Parsec.Char
-Text.Parsec.Combinator
-
 > import Text.Parsec (ParseError)
 > import Text.Parsec.String (Parser)
-
 > import Text.Parsec.String.Parsec (try, parse)
-
 > import Text.Parsec.String.Char (anyChar, oneOf, char, digit
 >                                ,string, letter, satisfy)
-
 > import Text.Parsec.String.Combinator (eof,manyTill, many1, choice
 >                                      ,anyToken,chainl1, between)
-
 > import Control.Applicative ((<$>), (<*>), (<$), (<*), (*>), (<|>), many)
 > import Control.Monad (void, ap)
 > import Data.Char (isLetter, isDigit)
@@ -169,7 +154,8 @@ This can be used to quickly try out new ad hoc parsers.
 = Type signatures
 
 TODO: update with new wrapper types which mean all this nonsense can
-be delayed to much later.
+be delayed to much later. Instead, a quick note here about the
+wrappers and how they help.
 
 I think you should always use type signatures with Parsec. Because the
 Parsec code is really generalized, without the type GHC will refuse to

Text/Parsec/String/Expr.hs

@@ -14,6 +14,8 @@ Wrappers for the Text.Parsec.Expr module with simplified types.
 import Text.Parsec.String (Parser)
 import qualified Text.Parsec.Expr as E
 
+-- not sure if this is neccessary, or a type alias would be good
+-- enough
 data Operator a = Infix (Parser (a -> a -> a)) E.Assoc
                 | Prefix (Parser (a -> a))
                 | Postfix (Parser (a -> a))

plan

@@ -16,12 +16,21 @@ review and remove explicit explanations where feasible, replace with
 more examples
 
 break up parsing introduction into several docs, it's huge
+1. IntroductionToParsing + Text.Parsec.Char review
+2. SimpleExpressionParser
+3. Text.Parsec.Combinators + Control.Applicative review
+4. Helper functions module + notes
 
 maybe have a sequence of docs for expression parsing, in the middle
 will be the simple sql value expression parser
 
 maybe split parsecextras file into three also?
 
+value expressions: start with the basic one in the first tutorial then
+have the one in the Expr tutorial then extend this with the simplified
+sql parser set of operators then create a new parser with each new
+ctor one at a time. This way can also do the naive thing first, then
+talk about how to deal with the problems that come up.
 
 documents which need finishing off: