Changeset 3144

Timestamp:

05/13/05 09:56:40 (4 years ago)

Author:

scook0

svk:copy_cache_prev:

4718

Message:

Updated Haddocks to reflect Autrijus's answers to my questions

Location:

src/Pugs

Files:

• AST/Internals.hs (modified) (3 diffs)
• Bind.hs (modified) (1 diff)
• Monads.hs (modified) (1 diff)

src/Pugs/AST/Internals.hs

@@ -620 +620 @@
 -- expressions ('Exp').
 type Bindings   = [(Param, Exp)]
+{-|
+A sub that has a non-empty 'SlurpLimit' is a bound (or partially bound) sub
+that has a finite number of slurpy scalar params bound, and no slurpy array
+param bound (see 'VCode' and "Pugs.Bind").
+
+Each list entry consists of the number of slurpable args expected, and an
+expression that will evaluate to the actual list of slurpable args.
+When the sub is called (see 'Pugs.Eval.apply'), the expression is evaluated.
+If it evaluates to /too many/ args, the call will fail.
+
+This needs to be a list (rather than a @Maybe@) because Perl6's @.assuming@
+(i.e. explicit currying) means that a sub can have its arguments bound in
+separate stages, and each of the bindings needs to be checked.
+
+>[12:02] <autrijus> scook0: .assuming will impose multiple limits
+>[12:02] <autrijus> because you can assume (curry) multiple times
+>[12:02] <scook0> ah
+>[12:02] <scook0> I'll have to write that in the docs then
+>[12:03] <scook0> Am I correct in that they only apply to subs that take a finite number of slurpy scalars?
+>[12:04] <scook0> Slurpy array params seem to nuke the SlurpLimit
+>[12:04] <scook0> because slurpy arrays can take any number of args
+>[12:07] <autrijus> scook0: yes, and yes.
+-}
 type SlurpLimit = [(VInt, Exp)]
 
@@ -869 +892 @@
 This is to allow for multi subs, because we will need to keep
 /multiple/ subs associated with one symbol. In other cases, the list
-should just contain a single value. See 'genSym' and 'genMultiSym' for
+should just contain a single value. See 'Pugs.AST.genSym' and 'Pugs.AST.genMultiSym' for
 more details.
 
@@ -880 +903 @@
 The current global and lexical pads are stored in the current 'Env', which
 is stored in the @Reader@-monad component of the current 'Eval' monad.
+
+>[11:56] <autrijus> scook0: I'm been thinking to split a Pad entry into single and multiple variants
+>[11:57] <autrijus> MkPad !(Map Var ([(TVar Bool, TVar VRef)]))
+>[11:57] <autrijus> becomes
+>[11:58] <autrijus> data Pad = MkPad !(Map Var PadEntry)
+>[11:58] <autrijus> data PadEntry = MkEntry (TVar Bool, TVar VRef) | MkEntryMulti [(TVar Bool, TVar VRef)]
+>[11:58] <autrijus> yeah. but it's not critical, so is low priority
 -}
 data Pad = MkPad !(Map Var ([(TVar Bool, TVar VRef)]))

src/Pugs/Bind.hs

@@ -49 +49 @@
 are given, only the first gets the arguments--the rest get an empty hash.
 Used by 'bindSomeParams'.
+
+>[12:16] <scook0> autrijus: At the moment, if you call a sub that has multiple slurpy arrays, 
+>                   Pugs deliberately binds the first one normally, and makes all the rest empty
+>[12:17] <scook0> Is this proper behaviour, or is it just a quirk of the current implementation?
+>[12:17] <autrijus> no, that's specced.
+>[12:17] <autrijus> i.e. correct
 -}
 bindHash :: [Exp]   -- ^ Named arguments (pair expressions) that were not

src/Pugs/Monads.hs

@@ -3 +3 @@
 {-|
     Monad structures.
+    
+    Note that there aren't actually any monads defined here--try looking in
+    "Pugs.AST.SIO" and "Pugs.AST.Internals".
 
 >   One Ring to rule them all,