ERRATA

	"Implementing functional languages: a tutorial"
	~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
	Peyton Jones & Lester, Prentice Hall 1992


Last modified: 10 Feb 94


*****************************************************************
*	Errata in VERSION 1.5 					*
*								*
* Version 1.5 corresponds to the second printing of the book	*
* You can distinguish it by looking at the very bottom of	*
* the fourth page of the book, which has the Libarary of	*
* Congress Cataloguing-in-Publication Data.			*
* At the bottom of this page, the second printing has a line	*
* which goes	2 3 4 5 96 95 94 93 92				*
* (The first printing has an extra "1" at the start 		*
*****************************************************************

Preface, page xii, line -16
	Replace "than a just a" by "than just a".

Fig 1.1, page 7.
	Interchange the last two lines of the table, so that
	underscore becomes a "varch" instead of an "alpha".

Section 1.5.4, page 17, exercise 1.6.
	Delete the phrase ", and define iDisplay in terms of flatten"
	(It already is so defined, a few lines above.)

Section 1.6.1, page 20, line -10 
	(Not an error; just an improvement.)
	Replace "123abc" by "123abc456"

Section 1.6.2, page 23, line 9: 
	"Exercise 1.16" should be "Exercise 1.17".

Section 1.6.2, page 23, line 12: 
	Replace "which which" with "which".

Section 1.6.4, page 27

    Replace 
	>          take_first_parse ((prog,[]) : other_parses) = prog
    by 
	>          take_first_parse ((prog,[]) : others) = prog
	>          take_first_parse (parse     : others) = take_first_parse others

    Insert "complete" just after "and then selecting the first", two
	sentences before this code.

    Replace the whole sentence immediately before the code by:
	If it returns no complete parse --- that is, one in which
	the sequence of tokens remaining after the parse is empty ---
	@syntax@ produces a (horribly uninformative) error message.


Section 1.6.6, page 30, Fig 1.3
	In line 4 of the Figure, replace "var" by "var_1 ... var_n"
	(The underbars means subscripts, of course)

Notes to Chapter 1, page 31, Note 4: 
	(Clarification, not error)
	Add "Such infix operators are right-associative".

Section 2.3.1, page 41, second-last bullet in list:
	Replace "the supercombinator named sc" by "a supercombinator".

Section 3.2.1, page 80, Exercise 3.1.
Section 3.2.1, page 80, Exercise 3.2.
	(Clarification.)
	Replace "for all expressions in aExpr" by 
		"for all expressions e of type aExpr"
		(both "e" and "aExpr" should be in typewriter font)
	Replace "aExpr" by "e" (twice) in the equation which follows.

Section 3.3.3, page 88, rule 3.12
Section 3.4.2, page 97, rule 3.17 (twice)
Section 3.5, page 99, rule 3.19
	(This isn't an error, but it makes the style more consistent.)
	Replace "Unwind : i" by "[Unwind]"

Setion 3.5.4, page 104, line -2
	Replace "an shorter" by "a shorter".

Section 3.7.2, page 116, line -9: 
	Replace "an simple" by "a simple".

Section 3.8.4, page 120, rule 3.31
	Replace "t:i'" by "t->i'"
	(the "->" should be in typewriter font)

Section 3.8.4, page 120, line -12, -13:
	Replace "1:" and "2:" by "1 ->" and "2 ->" respectively.
	The "Eval" on line -11 should be moved over to still line up
	with the "Split" above it as now.

Fig 3.14, page 122, lines 4 and 7
	In the two rules for Pack, add a superscript "+0" to the first 
	"rho" on the right hand side, and a superscript "+(a-1)" to the 
	second "rho" on the right hand side.

Fig 3.18, page 132, top line
	Replace "evalauates" by "evaluates"

Fig 3.20, page 134, line 14
	Replace "If" by "Cond"

Section 4.2.3, page 148, line -14
	After line -14 (which is the last line of the definition of compile),
	add the new line

	>	  ++ [(name, Label name) | (name, code) <- compliedPrimitives]

	The "++" should line up with the open square bracket of the previous
	line, to give

	>        initial_env = [(name, Label name) | (name, args, body) <- sc_defs]
	>	 	      ++ [(name, Label name) | (name, code) <- compliedPrimitives]

Section 4.2.4, page 150, line -16.
	add ", #stack >= n" to the end of the line
	insert the following new line just afterwards,
		> = error "Too few args for Take instruction",	otherwise
	The net effect is to make the following equation for step:

> step ((Take n:instr), fptr, stack, vstack, dump, heap, cstore,stats)
> = (instr, fptr', drop n stack, vstack, dump, heap', cstore, stats), #stack >= n
> = error "Too few args for Take instruction",			      otherwise
>   where (heap', fptr') = fAlloc heap (take n stack)

	(This change catches a runtime error, when Take is given a stack
	with too few arguments on it.  This can happen while updating
	is being added (especially before partial-application updating is in).
	The code as it stood originally silently ignored the error, giving
	rise to a much more obscure error later on.
	Furthermore, bugs in the compiler can easily give rise to an
	erroneous Take.)

Section 4.2.5, page 152, line -3
	Replace "# (hAddresses heap)" by "hSize heap"
	(This isn't a bug, but it's tidier.)

Figure 4.6, page 180, the E scheme
	In the right hand side of (d',is_body) = ...
	replace "R[e]" with "R[body]"

Exercise 4.28, page 185, line 9, replace "Take" by "UpdateMarkers"

Section 4.7.2, page 185, line -2, replace "timArg" by "timAMode"

Section 6.6.5, page 237, line 13
	replace
		> separateLams prog = [ (name, [], mkSepLams args rhs)
	by
		> separateLams prog = [ (name, [], mkSepLams args (separateLams_e rhs))

Section A.1.2, page 258, line -15
	Replace "a association" by "an association".

Section A.4.1, page 261, line -1
	Replace "foldl" by "foldll"

*****************************************************************
*	Errata in VERSION 1.2 					*
*								*
* Version 1.2 corresponds to the first printing of the book	*
* by Prentice Hall.						*
*****************************************************************

	If you find any other errors, 
	please email simonpj@dcs.glasgow.ac.uk


Ex 1.19.  "xxxxxx" should be ["x", "x", "x", "x", "x", "x"]

Section 2.6.2 "Implementing arithmetic", para after the 
	bullets describing primNeg: "If the stack contains just 
	one item, ..., and the dump is empty"
	
	Here "empty" should be "non-empty"

Section 2.7.1 "Building structured data"
	"NDatat" should be "NData"

Section 2.7.2 "Conditionals".
	if Pack{1,0} t e = t
	if Pack{2,0} t e = e

	The 1 and 2 should be interchanged.

Section 2.7.5 "Lists"
	In definition of caseList, swap the order of "cc" and "cn" on the 
	left-hand sides.  (cn should be before cc)

	In the following paragraph, replace "second" by "third" and
					    "third" by "second"

Section 2.8.3 "An alternative representation for data values"
	In the definitions of cons and nil near the end of the section,
	swap the order of "cc" and "cn" on the left-hand sides.
	(cn should be before cc)

Section 3.4.2 "The evaluator"
	In the rule 3.15 for Update, replace "a1" with "a0", twice.
	Add to the end of the sentence immediately before the rule 
		", and a0 is the function node."

Section 3.7.2 "The solution", second bullet
	Replace "e2" by "e1".

Section 4.2.3, page 148, line 23 (last line of "compile")
	Add a new line, and space out line 23 a little, to get
>        initial_env =    [(name, Label name) | (name, args, body) <- sc_defs]
>		       ++ [(name, Label name) | (name, code) <- compliedPrimitives]

Section 4.2.5, page 152, line -3
	replace "(# (hAddresses heap))" by "(hSize heap)"

Fig 4.2 (p160).   In the right-hand side of the B scheme equation for B[e1+e2],
	interchange e1 and e2, to read:

	B[e1+e2] rho cont = B[e2] rho (B[e1] rho (Op Add : cont))

Section 4.6.2 (p179) "Transition rules and compilation 
		      schemes for data structures".
	In Rule 4.19, on the second line of the rule, replace "f_d" by "f".
	In Rule 4.20, on the third line of the rule,  replace "f_d" by "f".

Fig 4.3 (p163), and Fig 4.7 (p187).  In the first equation for the R scheme,
	instead of:
		R[e] rho = B[e] rho [Return]
	the equation should read:
		R[e] rho d = B[e] rho d [Return]
		
Fig 4.4 (p172) and 4.7 (p187).  There should be a note in each Figure
	that the letrec case of the R scheme is identical to the let case,
	except that rho is replaced by rho' in each of the calls to U.

Fig 4.5 (p174) and 4.8 (p188).  Replace "Const" by "IntConst".
	[23 Oct 92]

Fig 4.6 (p180) and 4.7 (p187).  In the right-hand side of the R scheme for
	case expressions, replace "d2" by "d'".

Exercise 4.24 (p181).  Replace "Return" by "ReturnConstr t".

Fig 4.7 (p187).  In the right-hand side of the SC compilation scheme, replace
		Take d' n : is
	by
		UpdateMarkers n : Take d' n : is

Fig 4.8 (p188).  The B compilation scheme needs modification to incorporate
	depths.  The revised B scheme should read like this:

	B[e] rho d cont is a pair (d',is), where is is TIM code 
	which evaluates e in environment rho, putting its value 
	(which should be an integer) on top of the value stack,
	and continuing with the code sequence cont.
	The code assumes that the first d slots of the frame are 
	occupied, and it uses slots (d+1...d').

	B[e1 + e2] rho d cont = B[e2] rho d1 is1
				where (d1,is1) = B[e1] rho d (Op Add : cont)
	...and similar rules for other arithmetic primitives

	B[n] rho d cont = (d, PushV (IntVConst n) : cont)	where n is a number
	B[e] rho d cont = (d', Push (Code cont) : is)		otherwise
			  where (d',is) =  R[e] rho d


*****************************************************************
*								*
*	Errata in TUTOR's NOTES
*								*
*****************************************************************

gm-tutor.src, line 509: 
	[EConstr t 0] should be (EConstr t 0)


tim-tutor.src
	Make the same change to each definition of the step
	function as that described above for Section 4.2.4, page 150 above.

tim-tutor.src
	Make the same change to the definition of the compile function
	as that described above for Section 4.2.3, page 148, line -14