Naming and Logic
programming essentials with Wisp (WIP)
This article is a start of a text teaching programming basics with Wisp. It is a work in process. The tutorial starts from scratch and teaches in examples with minimal explanation. It’s core is a map of the minimal functionality you need from Scheme to write meaningful programs.
Currently it only contains entries for parts of the map. Each entry in the map will be fleshed out as I get to it. That may take weeks, months, or years, depending on whether and when it grips me.
I’d have loved to publish it only once all is written, but I considered it more important to get this into the open.
I hope that one day it will be interactive, using Guile on wasm (which is currently being built by the folks at Spritely).
To follow along, install Wisp and try the examples as you read.
The Map of Scheme
(How to make such a map? See Org Mode Tipp: imagemap)
Name a value: define
Use define to name a value. Use .
to return a value.
define small-tree-height-meters 3 define large-tree-height-meters 5 . small-tree-height-meters
After typing or copying a block into the Wisp REPL, hit enter three times. You should then see
$1 = 3
This means: the first returned value ($1
) is 3. The next time you
return a value, it will be called $2
.
Names can contain any letter except for (white-)space, quote, comma or parentheses. They must not be numbers.
define illegal name 1 define 'illegal-name 2 define ,illegal-name 3 define illegal)name 4 define 1113841 5
While compiling expression: Syntax error: unknown location: source expression failed to match any pattern in form (define illegal name 1) While reading expression: #<unknown port>:4:16: unexpected ")"
Use comparisons with numbers
= 3 5
$1 = #f
= 3 3
$1 = #t
#t
means true, #f
means false.
Returns the result of logic without needing a period (.
).
The logic comes first. This is clear for =
, but easy to misread for <
.
< 3 5 ;; is 3 smaller than 5? #true < 5 3 ;; is 5 smaller than 3? #false > 3 5 ;; is 3 bigger than 5? #false > 5 3 ;; is 5 bigger than 3? #true > 3 3 ;; is 3 bigger than 3? #false >= 3 3 ;; is 3 bigger or equal to 3? #true <= 3 3 ;; is 3 smaller or equal to 3? #true
$1 = #t $2 = #f $3 = #f $4 = #t $5 = #f $6 = #t $7 = #t
Use logic with infix
. {3 = 5} . {3 < 5} . {3 > 5}
$1 = #f $2 = #t $3 = #f
Infix logic gives a value, so you need .
to return it.
Because infix-logic gives a value, you can use it in place of a value.
For example to nest it:
. {{5 < 3} equal? #f}
$1 = #t
Or to name it as value:
define is-math-sane? {3 < 5} . is-math-sane?
$1 = #t
By convention, names that have the value true
or false
have the suffix ?
.
Use logic with names
define small-tree-height/m 3 define large-tree-height/m 5 . {small-tree-height/m < large-tree-height/m}
$1 = #t
Add comments with ;
define birch-height/m 3 ;; this is a comment define height ;; comment at the end ;; comment between lines . birch-height/m . height
$1 = 3
It is common to use ;;
instead of ;
, but not required.
Editors support this style.
Logic with true and false
and #t #t and #f #t or #f #t or #f #f
$1 = #t $2 = #f $3 = #t $4 = #f
If any value passed to and
is #f
(#false
), it ignores further values.
If any value passed to or
is not #f
(not #false
), it ignores further values.
and #t #t #t ;; => #true and #t #f #t ;; => #false and {3 < 5} {5 < 3} ;; => #false or #t #f #t ;; => true or {3 < 5} {5 < 3} ;; => #true or #f #f #f ;; => #false
$1 = #t $2 = #f $3 = #f $4 = #t $5 = #t $6 = #f
For and
and or
, everything is #true
(#t
) except for #false
(#f
).
Given the number of hard to trace errors in other languages that turn
up in production, this is the only sane policy.
Name the result of logic with indentation
define birch-h/m 3 define chestnut-h/m 5 define same-héight? = birch-h/m chestnut-h/m define smaller? . {birch-h/m < chestnut-h/m} . smaller?
$1 = #t
The infix directly gives a value, so it needs the .
as prefix to
return the value instead of trying to call the value as logic.
Name logic with define :
define : same-height? tree-height-a tree-height-b = tree-height-a tree-height-b same-height? 3 3
$1 = #t
By convention, logic that returns true or false has the suffix ?
.
You can now use your named logic like all other logic. Even with infix.
define : same-height? tree-height-a tree-height-b = tree-height-a tree-height-b . {3 same-height? 3}
$1 = #t
What this map of Scheme calls named logic is commonly called
function
or procedure
. We’ll stick with logic for the sake of
a leaner conceptual mapping.
The indented lines that define the named logic are called the body.
Name a name in define
with .
define small-tree-height-meters 3 define height . small-tree-height-meters . height
$1 = 3
.
returns the value of its line.
Return the value of logic with .
define : larger-than-4? size . {size > 4} . larger-than-4?
$1 = #<procedure larger-than-4? (size)>
The value of logic defined with define :
is a procedure
. You can
see the arguments in the output: If you call it with too few or too
many arguments, you get warnings.
There are other kinds of logic: syntax rules and reader-macros. We
will cover syntax rules later. New reader macros are rarely needed;
using {...}
for infix math is a reader macro.
Name in define :
with define
define birch-h/m 3 define : birch-is-small define reference-h/m 4 . { birch-h/m < reference-h/m } birch-is-small
$1 = #t
Only the last part is returned.
Note the .
in front of the { birch-h/m < reference-h/m }
: a
calculation inside braces is executed in-place, so its value needs to
be returned.
Name the result of logic in one line with :
or ()
define birch-h/m 3 define chestnut-h/m 5 define same-height : = birch-h/m chestnut-h/m . same-height define same-height-again (= birch-h/m chestnut-h/m) . same-height-again
$1 = #f $2 = #f
This is consistent with infix-math and uniform with defining logic:
define birch-h/m 3 define chestnut-h/m 5 define same-height {birch-h/m = chestnut-h/m} . same-height define : same-height? tree-height-a tree-height-b = tree-height-a tree-height-b define same-height : same-height? birch-h/m chestnut-h/m . same-height
$1 = #f $2 = #f
Name text with "
define tree-description "large tree" define footer "In Love, Arne" define greeting . "Hello" display footer
In Love, Arne
Like { }
, text (called string
as in “string of characters”) is its
value.
Text can span multiple lines. Line breaks in text do not affect code logic.
You can use \n
to add a line break within text without having a
visual line break. The backslash (\
) is the escape character and
\n
represents a line break. To type a real \
within quotes ( "
), you must escape it as \\
.
Return the value with .
to name text
with indentation.
With display
you can show text as it will look in an editor.
Text is stronger than comments:
define with-comment ;; belongs to coment ;; comment . "Hello ;; part of the text" . with-comment
$1 = "Hello ;; part of the text"
Take decisions with cond
define chestnut-h/m 5 define tree-description cond {chestnut-h/m > 4} . "large tree" : = 4 chestnut-h/m . "four meter tree" else . "small tree" . tree-description
$1 = "large tree"
cond
checks its clauses one by one and uses the first with value
#true
.
To cond
every valid value is #true
(#t
) except for #false
(#f
).
To use named logic, prefix it with :
to check its value.
cond 5 . #t else ;; else is #true in cond . #f cond #f . #f else . #t
$1 = #t $2 = #t
Use fine-grained numbers with number-literals
define more-precise-height 5.32517 define 100-meters 1e2 . more-precise-height . 100-meters
$1 = 5.32517 $2 = 100.0
These are floating point numbers. They store approximate values in 64 bit binary, depending on the platform. Read all the details in the Guile Reference manual Real and Rational Numbers, the r5rs numbers, and IEEE 754.
Use exact numbers with #e
and quotients
define exactly-1/5 #e0.2 define exactly-1/5-too 1/5 . exactly-1/5 . exactly-1/5-too
$1 = 1/5 $2 = 1/5
Guile computations with exact numbers stay reasonably fast even for unreasonably large or small numbers.
See inexact value of exact number with exact->inexact
exact->inexact #e0.2 exact->inexact 1/5 exact->inexact 2e7
$1 = 0.2 $2 = 0.2 $3 = 2.0e7
The inverse is inexact->exact
:
inexact->exact 0.5
$1 = 1/2
Note that a regular 0.2
need not be exactly 1/5
, because floating
point numbers do not have exact representation for that. You’ll need
#e
to have precise 0.2
.
inexact->exact 0.2 . #e0.2
$1 = 3602879701896397/18014398509481984 $2 = 1/5
Use math with the usual operators as logic
define one-hundred * 10 10 define half-hundred : / one-hundred 2 . half-hundred
$1 = 50
Remember that names cannot be valid numbers!
define 100 ;; error! * 10 10
While compiling expression: Syntax error: unknown location: source expression failed to match any pattern in form (define 100 (* 10 10))
Change the value or logic of a defined name with set!
define birch-h/m 3 set! birch-h/m 3.74 . birch-h/m
$1 = 3.74
It is customary to suffix named logic that changes values of existing names with !
.
Since logic can cause changes to names and not just return a result,
it is not called function
, but procedure
in documentation; proc
for brevity.
Create nameless logic with lambda
define : is-same-height? a b > a b ;; <- this is a mistake . is-same-height? is-same-height? 3 3 define : fixed a b = a b set! is-same-height? fixed . is-same-height? ;; but now called "fixed" in output! is-same-height? 3 3 ;; shorter and avoiding name pollution and confusion. set! is-same-height? lambda : a b = a b ;; must be on a new line ;; to not be part of the arguments. ;; since lambda has no name, we see the original name again . is-same-height? is-same-height? 3 3
$1 = #<procedure is-same-height? (a b)> $2 = #f $3 = #<procedure fixed (a b)> $4 = #t $5 = #<procedure is-same-height? (a b)> $6 = #t
The return value of lambda
is logic (a procedure
).
Logic knows the name it has been defined as, except if it is defined
via lambda
.
lambda
is a special form. Think of it as define : name arguments
,
but without the name.
Return list of values with list
list 3 5 define known-heights list 3 3.75 5 100 . known-heights
$1 = (3 5) $2 = (3 3.75 5 100)
You can put values on their own lines by returning their value: .
returns all the values in its line. Different from define :
, list keeps
all values, not just the last.
define known-heights-2 list 3 . 3.75 5 . 100 define known-heights-3 list . 3 . 3.75 . 5 . 100 define : last-height . 3 3.75 5 100 = 100 : last-height
$1 = #t
Compare structural values with equal?
= 3 3 3 ;; reuse name definition snippets {{{known-heights}}} {{{known-heights2}}} equal? known-heights known-heights-2 known-heights-3
$1 = #t $2 = (3 5) $3 = (3 3.75 5 100) $4 = #t $5 = #t
Like =
and +
, equal?
can be used on arbitrary numbers of values.
Reusing the snippets here uses the noweb
syntax via Emacs Org Mode.
Apply logic to a list of values with apply
apply = : list 3 3 equal? = 3 3 apply = list 3 3
$1 = #t $2 = #t
Only the last argument of apply is treated as list, so you can give initial arguments:
define a 1 define b 1 apply = a b list 1 1
$1 = #t
Get the arguments of named logic as list with . args
define : same? heights apply = heights same? : list 1 1 1 same? list 1 1 1 define : same2? . heights apply = heights same2? 1 1 1 same2? . 1 1 1
$1 = #t $2 = #t $3 = #t $4 = #t
These are called rest
. Getting them is not for efficiency: the list creation is implicit.
You can mix regular arguments and rest
arguments:
define : same? alice bob . rest display : list alice bob rest newline apply = alice bob rest same? 1 1 1 1
(1 1 (1 1)) $1 = #t
Remember that apply uses only the last of its arguments as list, in symmetry with . rest
.
Get the result of applying logic to each value in lists with map
define birch-h/m 3 define : same-height-as-birch? height/m = birch-h/m height/m define heights : list 3 3.75 5 100 . heights map same-height-as-birch? . heights map + list 1 2 3 list 3 2 1 map list list 1 2 3 list 3 2 1
$1 = (3 3.75 5 100) $2 = (#t #f #f #f) $3 = (4 4 4) $4 = ((1 3) (2 2) (3 1))
When operating on multiple lists, it takes one argument from each list. All lists must be the same length.
To remember: apply
extracts the values from its last argument, map
extracts one value from each argument after the first.
Apply logic to each value in lists and ignoring the results with for-each
define birch-h/m 3 define has-birch-height #f define : set-true-if-birch-height! height/m cond {birch-h/m = height/m} set! has-birch-height #t define heights : list 3 3.75 5 100 for-each set-true-if-birch-height! heights . has-birch-height
$1 = #t
Import pre-defined named logic and values with import
import : ice-9 pretty-print srfi :1 lists ;; no space after :. it is part of the name pretty-print list 12 list 34 . 5 6 first : list 1 2 3 ;; 1 second : list 1 2 3 ;; 2 third : list 1 2 3 ;; 3 member 2 : list 1 2 3 ;; list 2 3 => #true
(12 (34) 5 6) $1 = 1 $2 = 2 $3 = 3 $4 = (2 3)
Import uses modules which can have multiple components. In the first
import, ice-9
is one component and the second is pretty-print
. In
the second, srfi
is the first component, :1
is the second, and
lists
is the third.
ice-9 is the name for the core extensions of Guile. It’s a play on ice-nine, a fictional perfect seed crystal.
SRFI’s are Scheme Requests For Implementation, portable libraries
built in collaboration between different Scheme implementations. The
ones available in Guile can be found in the Guile reference manual.
More can be found on srfi.schemers.org. They are imported by number
(:1
) and can have a third component with a name, but that’s not
required.
You can use only
to import only specific names.
import : only (srfi :1) first second only (srfi :1) iota first : list 1 2 3 ;; 1 second : list 1 2 3 ;; 2 third : list 1 2 3 ;; error iota 5 ;; list 0 1 2 3 4
$1 = 1
$2 = 2
ice-9/boot-9.scm:1683:22: In procedure raise-exception:
Unbound variable: third
Entering a new prompt. Type `,bt' for a backtrace or `,q' to continue.
$3 = (0 1 2 3 4)
Apply partial procedures with srfi :26 cut
define : plus-3 number + 3 number map plus-3 list 1 2 3 ;; list 4 5 6 import : srfi :26 cut map : cut + 3 <> list 1 2 3 ;; list 4 5 6 map : cut - <> 1 ;; => <> - 1 list 1 2 3 ;; list 0 1 2 map : cut - 1 <> ;; => 1 - <> list 1 2 3 ;; list 0 -1 -2 define plus-3-cut : cut + 3 <> map plus-3-cut list 1 2 3 ;; list 4 5 6
$1 = (4 5 6) $2 = (4 5 6) $3 = (0 1 2) $4 = (0 -1 -2) $5 = (4 5 6)
Use r7rs
datatypes, e.g. with vector-map
R7RS is the 7th Revised Report
on Scheme. Guile provides a superset of the standard: its core can be
imported as scheme base
. A foundational datatype is Vectors with
O(1)
random access guarantee.
import : scheme base define vec : list->vector '(1 b "third") vector-map : λ (element) : cons 'el element . vec
$1 = #((el . 1) (el . b) (el . "third"))
Vectors have the literal form #(a b c)
. It is an error to mutate it.
import : scheme base define mutable-vector : list->vector '(1 b "third") define literal-vector #(1 b "third") vector-set! mutable-vector 1 "b" ;; allowed . mutable-vector vector-set! literal-vector 1 "b" ;; forbidden . literal-vector
$1 = #(1 "b" "third")
ice-9/boot-9.scm:1683:22: In procedure raise-exception:
In procedure vector-set!: Wrong type argument in position 1 (expecting mutable vector): #(1 b "third")
Entering a new prompt. Type `,bt' for a backtrace or `,q' to continue.
$2 = #(1 b "third")
Name structured values with define-record-type
import : srfi :9 records define-record-type <tree> make-tree type height-m weight-kg carbon-kg . tree? type tree-type height-m tree-height weight-kg tree-weight carbon-kg tree-carbon define birch-young make-tree "birch" 13 90 45 ;; 10 year, 10cm diameter, define birch-old make-tree "birch" 30 5301 2650 ;; 50 year, 50cm define birch-weights map tree-weight : list birch-young birch-old . birch-young . birch-old . birch-weights
$1 = #<<tree> type: "birch" height-m: 13 weight-kg: 90 carbon-kg: 45> $2 = #<<tree> type: "birch" height-m: 30 weight-kg: 5301 carbon-kg: 2650> $3 = (90 5301)
Carbon content in birch trees is about 46% to 50.6% of the mass. See forestry commission technical paper 1993.
Height from Waldwissen, weight from BaumUndErde.
Get the result of logic inline with parentheses ()
, braces {}
, or colon :
import : srfi :26 cut list 1 2 : + 1 2 . 4 list 1 2 {1 + 2} 4 list 1 2 (+ 1 2) 4 map (cut + 3 <>) : list 1 2 3
$1 = (1 2 3 4) $2 = (1 2 3 4) $3 = (1 2 3 4) $4 = (4 5 6)
Line breaks and indentation are ignored inside parentheses, except for
the value of strings
.
The operators that need linebreaks are disabled: colon :
and period
.
do not get the value or return it, but the last value is returned
implicitly. This is the default in regular Scheme.
:
needs linebreaks, because it only goes to the end of the line.
.
needs linebreaks, because it only applies at the beginning of the
line (after indentation).
cut
is logic that has logic as result.
Handle errors with-exception-handler
;; unhandled exception stops execution define : add-5 input + 5 input ;; illegal for text map add-5 ' : "five" 6 "seven" ;; check inputs define : add-5-if input if : number? input + 5 input . #f map add-5-if ' : "five" 6 "seven" ;; handle exceptions define : add-5-handler input with-exception-handler λ (e) : format #t "must be number, is ~S.\n" input . #f λ () : + 5 input . #:unwind? #t ;; #t: continue #f: stop map add-5-handler ' : "five" 6 "seven"
$1 = #f ice-9/boot-9.scm:1683:22: In procedure raise-exception: In procedure +: Wrong type argument in position 1: "five" Entering a new prompt. Type `,bt' for a backtrace or `,q' to continue. $2 = (#f 11 #f) must be number, is "five". must be number, is "seven". $3 = (#f 11 #f)
In Guile Wisp checking inputs is often cheaper than exception handling.
Test your code with srfi 64
Is your code correct?
import : srfi :64 testsuite define : tree-carbon weight-kg * 0.5 weight-kg define : run-tests test-begin "test-tree-carbon" test-equal 45.0 tree-carbon 90 test-approximate 45.0 + 40 : random 10.0 . 5 ;; expected error size test-assert : equal? 45.0 : tree-carbon 90 test-error : throw 'wrong-value test-end "test-tree-carbon" run-tests
%%%% Starting test test-tree-carbon (Writing full log to "test-tree-carbon.log") # of expected passes 4 $1 = #<test-runner pass-count: 4 fail-count: 0 xpass-count: 0 xfail-count: 0 skip-count: 0 skip-list: () fail-list: () run-list: #t skip-save: () fail-save: () group-stack: () on-test-begin: #<procedure test-on-test-begin-simple (runner)> on-test-end: #<procedure test-on-test-end-simple (runner)> on-group-begin: #<procedure test-on-group-begin-simple (runner suite-name count)> on-group-end: #<procedure test-on-group-end-simple (runner)> on-final: #<procedure 70e68a95b240 at srfi/srfi-64/testing.scm:285:12 (r)> on-bad-count: #<procedure test-on-bad-count-simple (runner count expected-count)> on-bad-end-name: #<procedure test-on-bad-end-name-simple (runner begin-name end-name)> total-count: 4 count-list: () result-alist: ((source-form test-end "test-tree-carbon") (source-line . 15)) aux-value: #<output: test-tree-carbon.log 13>>
You can use this anywhere.
For details, see srfi 64.
Define derived logic structures with define-syntax-rule
In usual logic application in procedures
, arguments are
evaluated to their return value first.
Procedures
evaluate from inside to outside:
import : ice-9 pretty-print define : hello-printer . args pretty-print "Hello" for-each pretty-print args hello-printer 1 : pretty-print "second" . 3 4 ;; prints "second" "Hello" 1 3 4
"second" "Hello" 1 #<unspecified> 3 4
But for example cond
only evaluates the required branches. It is not
a procedure
, but a syntax-rule
.
Syntax-rules evaluate from outside to inside:
import : ice-9 pretty-print define-syntax-rule : early-printer args ... begin pretty-print "Hello" for-each pretty-print : list args ... early-printer 1 : pretty-print "second" . 3 4 ;; prints "Hello" "second" 1 3 4
"Hello" "second" 1 #<unspecified> 3 4
Arguments of define-syntax-rule
are only evaluated when they are
passed into a regular procedure
or returned. By calling other
syntax-rules
in syntax-rules
, evaluation can be delayed further.
define-syntax-rule
can reorder arguments and pass them to
procedures
and other syntax-rules
. It cannot ask for argument
values, because it does not evaluate names as values. It operates on
names and structure.
Instead of define : name . args
, it uses a pattern
define-syntax-rule : name args ...
The ellipsis ...
marks args
as standing for zero or more names. It
must be used with the ellipsis.
The body of define-syntax-rule
must only have one element. The logic
begin
wraps its own body to count as only one element. It returns
the value of the last element in its body.
Get and resolve names used in code with quote
, eval
, and module-ref
list : quote alice quote bob quote carol quote dave ;; => (alice bob carol dave) define alice "the first" eval 'alice : current-module ;; => "the first" module-ref (current-module) 'alice ;; => "the first" ;; module-ref is less powerful than eval. And safer. define code quote list 1 2 3 . code ;; => (list 1 2 3) ;; uses parentheses form eval code : current-module ;; => (1 2 3) ' 1 2 3 ;; (1 2 3) list 1 2 3 ;; (1 2 3) equal? : ' 1 2 3 list 1 2 3
$1 = (alice bob carol dave) $2 = "the first" $3 = "the first" $4 = (list 1 2 3) $5 = (1 2 3) $6 = (1 2 3) $7 = (1 2 3) $8 = #t
The form ' 1 2 3
is a shorthand to create an immutable
(literal) list that is equal?
to list 1 2 3
.
But some operations like list-set! the-list index new-value
from srfi :1
do not work on immutable lists.
define mutable-list : list 1 2 3 list-set! mutable-list 1 'a ;; zero-indexed . mutable-list define immutable-list : ' 1 2 3 . immutable-list list-set! immutable-list 1 'a ;; error!
$1 = a
$2 = (1 a 3)
$3 = (1 2 3)
ice-9/boot-9.scm:1683:22: In procedure raise-exception:
In procedure set-car!: Wrong type argument in position 1 (expecting mutable pair): (2 3)
Entering a new prompt. Type `,bt' for a backtrace or `,q' to continue.
Build value-lists with quasiquote
and unquote
define : tree-manual type height weight carbon-content list : cons 'type type cons 'height height cons 'weight weight cons 'carbon-content carbon-content tree-manual "birch" 13 90 45 define : tree-quasiquote type height weight carbon-content quasiquote : type . : unquote type height . : unquote height weight . : unquote weight carbon-content . : unquote carbon-content tree-quasiquote "birch" 13 90 45 define : tree-shorthand type height weight carbon-content ` : type . ,type ;; ` is short for quasiquoted list height . ,height ;; , is short for unquote weight . ,weight carbon-content . ,carbon-content tree-shorthand "birch" 13 90 45
$1 = ((type . "birch") (height . 13) (weight . 90) (carbon-content . 45)) $2 = ((type . "birch") (height . 13) (weight . 90) (carbon-content . 45)) $3 = ((type . "birch") (height . 13) (weight . 90) (carbon-content . 45))
These three methods are almost equivalent, except that quasiquoting can create an immutable list, but this is not guaranteed.
define three 3 define mutable-list : list 1 2 3 list-set! mutable-list 1 'a ;; zero-indexed . mutable-list define immutable-list : ` 1 2 3 list-set! immutable-list 1 'a ;; error! . immutable-list define mutable-quasiquoted : ` 1 2 ,three list-set! mutable-quasiquoted 1 'a ;; currently no error! . mutable-quasiquoted
$1 = a
$2 = (1 a 3)
ice-9/boot-9.scm:1683:22: In procedure raise-exception:
In procedure set-car!: Wrong type argument in position 1 (expecting mutable pair): (2 3)
Entering a new prompt. Type `,bt' for a backtrace or `,q' to continue.
$3 = (1 2 3)
$4 = a
$5 = (1 a 3)
Mutating quasiquoted lists may throw an error in the future. From the standard:
A quasiquote expression may return either newly allocated, mutable objects or literal structure for any structure that is constructed at run time …
Reuse your logic with let-recursion
Remember the for-each
example:
define has-birch-height #f define : set-true-if-birch-height! height/m define birch-h/m 3 cond {birch-h/m = height/m} set! has-birch-height #t define heights : list 3 3.75 5 100 for-each set-true-if-birch-height! heights . has-birch-height
$1 = #t
Instead of for-each
, we can build our own iteration:
define : has-birch-height? heights define birch-h/m 3 let loop : : heights heights cond (null? heights) #f : = birch-h/m : car heights ;; car is first . #t else loop : cdr heights define heights : list 3 3.75 5 100 has-birch-height? heights
$1 = #t
null?
asks whether the list is empty. car
gets the first element
of a list, cdr
gets the list without its first element.
Recursion is usually easier to debug (all variable elements are available at the top of the let recursion) and often creates cleaner APIs than iteration.
As rule of thumb: start with the recursion end condition (here:
(null? heights)
and ensure that each branch of the cond
either
ends recursion or moves a step towards finishing (usually with cdr
).
Another example why recursion wins:
define : fib n let rek : (i 0) (u 1) (v 1) if : >= i : - n 2 . v rek (+ i 1) v (+ u v)
Extend a list with cons
To build your own map
function which returns a list of results, you
need to add to a list.
cons 1 : list 2 3 ;; => list 1 2 3
$1 = (1 2 3)
Used for a simplified map
implementation that accepts only a single list:
import : only (srfi :1) first define : single-map proc elements let loop : (changed (list)) (elements elements) cond : null? elements reverse changed else loop ;; add processed first element to changed cons : proc : first elements . changed ;; drop first element from elements cdr elements single-map even? : list 1 2 3 ;; => #f #t #f
$1 = (#f #t #f)
Document procedures with docstrings
define : documented-proc arg . "Proc is documented" . #f ;; doc must not be last element procedure-documentation documented-proc ;; variables have no docstrings but ;; properties can be set manually. define variable #f set-object-property! variable 'documentation . "Variable is documented" object-property variable 'documentation
$1 = "Proc is documented" $2 = "Variable is documented" $3 = "Variable is documented"
You can get the documentation with help
or ,d
on the REPL:
,d documented-proc => Proc is documented ,d variable => Variable is documented
For generating documentation from comments, there’s guild doc-snarf
.
;; Proc docs can be snarfed define : snarfed-proc arg . #f ;; Variable docs can be snarfed define snarfed-variable #f
If this is saved as hello.w, get the docs via
wisp2lisp hello.w > hello.scm && \
guild doc-snarf --texinfo hello.scm
Read the docs
Now you understand the heart of code. With this as the core there is one more step, the lifeblood of programming: learning more. Sources:
- the Guile Reference manual
- Scheme Requests for Implementation (SRFI): tagged libraries
- The Scheme standards (RnRS), specifically r7rs-small (pdf)
- a list of tools and libraries
- Rosetta Code with solutions to many algorithm problems
Info manuals can often be read online, but the info
commandline
application and info
in Emacs (C-h i
) are far more efficient and
provide full-text search. You can use it to read the Guile reference
manual and some libraries. Get one by installing texinfo or Emacs.
In interactive wisp
(the REPL), you can check documentation:
help string-append .
`string-append' is a procedure in the (guile) module. - Scheme Procedure: string-append . args Return a newly allocated string whose characters form the concatenation of the given strings, ARGS.
,help
Help Commands [abbrev]: ...
Create a manual as package documentation with texinfo
Create a doc/
folder and add a hello.texi
file.
An example file can look like the following:
@documentencoding UTF-8 @settitle Hello World @c This is a comment @c The Top node is the first page for info @node Top @c Show the title @top @menu * First Steps:: * API Reference:: @end menu @contents @node First Steps @chapter First Steps Quick start: install, minimal example @itemize @item Download from ... @item Install: @code{make}. @end itemize Example: @lisp (+ 1 2) @result{} 3 @end lisp @node API Reference @chapter API Reference @section Procedures @subsection hello Print Hello @example hello @end example
Add a Makefile
in the doc/ folder:
all: hello.info hello.epub hello_html/index.html hello.info: hello.texi makeinfo hello.texi hello.epub: hello.texi makeinfo --epub hello.texi hello_html/index.html: hello.texi makeinfo --html hello.texi
Run make:
make
Read the docs with calibre
or the browser
or plain info
:
calibre hello.epub & \ firefox hello_html/index.html & \ info -f ./hello.info
The HTML output is plain. You can adapt it with CSS by adding
--css-include=FILENAME
or --css-ref=URL
.
Alternately you can write an Org Mode document and evaluate
(require 'ox-texinfo)
to activate exporting to texinfo.
Track changes with a version tracking system like Mercurial
or Git
For convenience, first initialize a version tracking repository, for example Mercurial or Git.
# either Mercurial hg init hello # or Git git init hello # enter the repository folder cd hello/
Now you can add new files with
# in Mercurial hg add FILE # in Git git add FILE
And take a snapshot of changes with
# in Mercurial hg commit -m "a change description" # in Git git commit -a -m "a change description"
It is good practice to always use a version tracking system.
For additional information and how to publish your code if you want to, see the Mercurial Guide or the Git Tutorial.
Package your code with autoconf
and automake
Create a configure.ac
file with name, contact info and version.
dnl configure.ac dnl Name, Version, and contact information. AC_INIT([hello], [0.0.1], [myName@example.com]) # Find a supported Guile version and set it as @GUILE@ GUILE_PKG([3.0]) GUILE_PROGS GUILE_SITE_DIR AC_PREFIX_PROGRAM([guile]) AM_INIT_AUTOMAKE([gnu]) AC_CONFIG_FILES([Makefile]) AC_OUTPUT
Add a Makefile.am
with build rules. Only the start needs to be edited:
## Makefile.am rules start bin_SCRIPTS = hello # program name SUFFIXES = .w .scm .sh WISP = hello.w # source files hello: $(WISP) echo "#!/usr/bin/env bash" > "$@" && \ echo "exec -a $$0 guile --language=wisp -s \"$@\"" \ >> "$@" && \ echo ";; exec block finished here: !#" >> "$@" && \ cat "$^" >> "$@" && chmod +x "$@" # integrate the texinfo documentation hello_TEXINFOS = doc/hello.texi # add library files, prefix nobase_ preserves directories nobase_site_DATA =
The rest of the Makefile.am can be copied verbatim.
## Makefile.am technical details # where to install guile modules to import. See # https://www.gnu.org/software/automake/manual/html_node/Alternative.html sitedir = $(datarootdir)/guile/site/$(GUILE_EFFECTIVE_VERSION) GOBJECTS = $(nobase_site_DATA:%.w=%.go) nobase_go_DATA = $(GOBJECTS) godir=$(libdir)/guile/$(GUILE_EFFECTIVE_VERSION)/site-ccachep # Make sure that the mtime of installed compiled files # is greater than that of installed source files. See: # http://lists.gnu.org/archive/html/guile-devel/2010-07/msg00125.html # The missing underscore before DATA is intentional. guile_install_go_files = install-nobase_goDATA $(guile_install_go_files): install-nobase_siteDATA EXTRA_DIST = $(WISP) CLEANFILES = $(GOBJECTS) $(wildcard *~) DISTCLEANFILES = $(bin_SCRIPTS) $(nobase_site_DATA) # precompile all source files .w.go: $(GUILE_TOOLS) compile --from=wisp $(GUILE_WARNINGS) \ -o "$@" "$<"
## Makefile.am help .PHONY: help help: ## Show this help message. @echo 'Usage:' @echo ':make [target] ...' \ | sed "s/\(target\)/\\x1b[36m\1\\x1b[m/" \ | column -c2 -t -s : @echo @echo 'Custom targets:' @echo -e "$$(grep -hE '^\S+:.*##' $(MAKEFILE_LIST) \ | sed -e \ 's/:.*##\s*/:/' -e \ 's/^\(.\+\):\(.*\)/:\\x1b[36m\1\\x1b[m:\2/' \ | column -c2 -t -s :)" @echo @echo '(see ./configure --help for setup options)'
This assumes that the folder hello
uses a Version tracking system.
## Makefile.am basic additional files .SECONDARY: ChangeLog AUTHORS ChangeLog: ## create the ChangeLog from the history echo "For user-visible changes, see the NEWS file" > "$@" echo >> "$@" if test -d "/home/arne/Schreibtisch/arnebab-org/.git"; \ then cd /home/arne/Schreibtisch/arnebab-org \ && git log --date-order --date=short \ | sed -e '/^commit.*$/d' \ | awk '/^Author/ {sub(/\\$/,""); getline t; print $0 t; next}; 1' \ | sed -e 's/^Author: //g' \ | sed -e \ 's/\(.*\)>Date: \([0-9]*-[0-9]*-[0-9]*\)/\2 \1>/g' \ | sed -e 's/^\(.*\) \(\)\t\(.*\)/\3 \1 \2/g' \ >> "$@"; cd -; fi if test -d "/home/arne/Schreibtisch/arnebab-org/.hg"; \ then hg -R "/home/arne/Schreibtisch/arnebab-org" \ log --style changelog \ >> "$@"; fi AUTHORS: ## create the AUTHORS file from the history if test -d "/home/arne/Schreibtisch/arnebab-org/.git"; \ then cd "/home/arne/Schreibtisch/arnebab-org" \ && git log --format='%aN' \ | sort -u >> "$@"; cd -; fi if test -d "/home/arne/Schreibtisch/arnebab-org/.hg"; \ then hg -R "/home/arne/Schreibtisch/arnebab-org" \ --config extensions.churn= \ churn -t "{author}" >> "$@"; fi
Now create a README
and a NEWS
file:
#+title: Hello A simple example project. * Requirements - Guile version 3.0.10 or later. * Build the project #+begin_src bash ,# Build the project autoreconf -i && ./configure && make ,# Create a distribution tarball autoreconf -i && ./configure && make dist #+end_src * License GPLv3 or later.
hello 0.0.1 - initialized the project
And for the sake of this example, a simple hello.w
file:
display "Hello World!\n"
To Be Continued …
Also see the Map of R7RS and the Scheme primer.