Capturing Macros in Scheme - macros

What's the simplest way to define a capturing macro using define-syntax or define-syntax-rule in Racket?
As a concrete example, here's the trivial aif in a CL-style macro system.
(defmacro aif (test if-true &optional if-false)
`(let ((it ,test))
(if it ,if-true ,if-false)))
The idea is that it will be bound to the result of test in the if-true and if-false clauses. The naive transliteration (minus optional alternative) is
(define-syntax-rule (aif test if-true if-false)
(let ((it test))
(if it if-true if-false)))
which evaluates without complaint, but errors if you try to use it in the clauses:
> (aif "Something" (displayln it) (displayln "Nope")))
reference to undefined identifier: it
The anaphora egg implements aif as
(define-syntax aif
(ir-macro-transformer
(lambda (form inject compare?)
(let ((it (inject 'it)))
(let ((test (cadr form))
(consequent (caddr form))
(alternative (cdddr form)))
(if (null? alternative)
`(let ((,it ,test))
(if ,it ,consequent))
`(let ((,it ,test))
(if ,it ,consequent ,(car alternative)))))))))
but Racket doesn't seem to have ir-macro-transformer defined or documented.

Racket macros are designed to avoid capture by default. When you use define-syntax-rule it will respect lexical scope.
When you want to "break hygiene" intentionally, traditionally in Scheme you have to use syntax-case and (carefully) use datum->syntax.
But in Racket the easiest and safest way to do "anaphoric" macros is with a syntax parameter and the simple define-syntax-rule.
For example:
(require racket/stxparam)
(define-syntax-parameter it
(lambda (stx)
(raise-syntax-error (syntax-e stx) "can only be used inside aif")))
(define-syntax-rule (aif condition true-expr false-expr)
(let ([tmp condition])
(if tmp
(syntax-parameterize ([it (make-rename-transformer #'tmp)])
true-expr)
false-expr)))
I wrote about syntax parameters here and also you should read Eli Barzilay's Dirty Looking Hygiene blog post and Keeping it Clean with Syntax Parameters paper (PDF).

See Greg Hendershott's macro tutorial. This section uses anaphoric if as example:
http://www.greghendershott.com/fear-of-macros/Syntax_parameters.html

Although the answer above is the accepted way to implement aif in the Racket community, it has severe flaws. Specifically, you can shadow it by defining a local variable named it.
(let ((it 'gets-in-the-way))
(aif 'what-i-intended
(display it)))
The above would display gets-in-the-way instead of what-i-intended, even though aif is defining its own variable named it. The outer let form renders aif's inner let definition invisible. This is what the Scheme community wants to happen. In fact, they want you to write code that behaves like this so badly, that they voted to have my original answer deleted when I wouldn't concede that their way was better.
There is no bug-free way to write capturing macros in Scheme. The closest you can come is to walk down the syntax tree that may contain variables you want to capture and explicitly strip the scoping information that they contain, replacing it with new scoping information that forces them to refer to your local versions of those variables. I wrote three "for-syntax" functions and a macro to help with this:
(begin-for-syntax
(define (contains? atom stx-list)
(syntax-case stx-list ()
(() #f)
((var . rest-vars)
(if (eq? (syntax->datum #'var)
(syntax->datum atom))
#t
(contains? atom #'rest-vars)))))
(define (strip stx vars hd)
(if (contains? hd vars)
(datum->syntax stx
(syntax->datum hd))
hd))
(define (capture stx vars body)
(syntax-case body ()
(() #'())
(((subform . tl) . rest)
#`(#,(capture stx vars #'(subform . tl)) . #,(capture stx vars #'rest)))
((hd . tl)
#`(#,(strip stx vars #'hd) . #,(capture stx vars #'tl)))
(tl (strip stx vars #'tl)))))
(define-syntax capture-vars
(λ (stx)
(syntax-case stx ()
((_ (vars ...) . body)
#`(begin . #,(capture #'(vars ...) #'(vars ...) #'body))))))
That gives you the capture-vars macro, which allows you to explicitly name the variables from the body you'd like to capture. aif can then be written like this:
(define-syntax aif
(syntax-rules ()
((_ something true false)
(capture-vars (it)
(let ((it something))
(if it true false))))
((_ something true)
(aif something true (void)))))
Note that the aif I have defined works like regular Scheme's if in that the else-clause is optional.
Unlike the answer above, it is truly captured. It's not merely a global variable:
(let ((it 'gets-in-the-way))
(aif 'what-i-intended
(display it)))
The inadequacy of just using a single call to datum->syntax
Some people think that all you have to do to create a capturing macro is use datum->syntax on one of the top forms passed to your macro, like this:
(define-syntax aif
(λ (stx)
(syntax-case stx ()
((_ expr true-expr false-expr)
(with-syntax
((it (datum->syntax #'expr 'it)))
#'(let ((it expr))
(if it true-expr false-expr))))
((_ expr true-expr)
#'(aif expr true-expr (void))))))
Just using datum->syntax is only a 90% solution to writing capturing macros. It will work in most cases, but break in some cases, specifically if you incorporate a capturing macro written this way in another macro. The above macro will only capture it if the expr comes from the same scope as the true-expr. If they come from different scopes (this can happen merely by wrapping the user's expr in a form generated by your macro), then the it in true-expr will not be captured and you'll be left asking yourself "WTF won't it capture?"
You may be tempted to quick-fix this by using (datum->syntax #'true-expr 'it) instead of (datum->syntax #'expr 'it). In fact this makes the problem worse, since now you won't be able to use aif to define acond:
(define-syntax acond
(syntax-rules (else)
((_) (void))
((_ (condition . body) (else . else-body))
(aif condition (begin . body) (begin . else-body)))
((_ (condition . body) . rest)
(aif condition (begin . body) (acond . rest)))))
If aif is defined using the capture-vars macro, the above will work as expected. But if it's defined by using datum->syntax on the true-expr, the the addition of begin to the bodies will result in it being visible in the scope of acond's macro definition instead of the code that invoked acond.
The impossibility of really writing a capturing macro in Racket
This example was brought to my attention, and demonstrates why you just can't write a real capturing macro in Scheme:
(define-syntax alias-it
(syntax-rules ()
((_ new-it . body)
(let ((it new-it)) . body))))
(aif (+ 1 2) (alias-it foo ...))
capture-vars cannot capture the it in alias-it's macroexpansion, because it won't be on the AST until after aif is finished expanding.
It is not possible at all to fix this problem, because the macro definition of alias-it is most probably not visible from the scope of aif's macro definition. So when you attempt to expand it within aif, perhaps by using expand, alias-it will be treated as a function. Testing shows that the lexical information attached to alias-it does not cause it to be recognized as a macro for a macro definition written out of scope from alias-it.
Some would argue that this shows why the syntax-parameter solution is the superior solution, but perhaps what it really shows is why writing your code in Common Lisp is the superior solution.

Related

Racket - How to define a function that can be used both in syntax transformers and ordinary code?

I am using syntax transformers to define macros in Racket. I want to create some helper functions to help me manipulate the syntax. However, the functions I defined outside the syntax transformer are not available inside the syntax transformer. For example, in the following code
(define (my-function x) (+ x 1))
(define-syntax my-macro
(lambda (stx)
(datum->syntax stx (my-function (cadr (syntax->datum stx))))))
I got the error "my-function: reference to an unbound identifier at phase: 1; the transformer environment".
After some searching, I am able to write the following code so that my-function is available inside the syntax transformer.
(begin-for-syntax
(define (my-function x) (+ x 1)))
(provide (for-syntax my-function))
(define-syntax my-macro
(lambda (stx)
(datum->syntax stx (my-function (cadr (syntax->datum stx))))))
But the problem is, my-function is not available outside the syntax transformer this time. Sometimes I want to check those helper functions in ordinary code, so I need to be able to call it from both inside and outside the syntax transformer, just like the function cadr. How can I achieve that?
I know my question has something to do with Racket's syntax model, in particular the concept of "phase level", but I never really understand it. If you could provide some easy-to-follow tutorials explaining it I would even be more grateful.
A common way is to define your function that you want to share across phases in another (sub)module. Then, require it twice.
#lang racket
(module common racket
(provide my-function)
(define (my-function x) (+ x 1)))
(require 'common
(for-syntax 'common))
(define-syntax my-macro
(lambda (stx)
(datum->syntax stx (my-function (cadr (syntax->datum stx))))))
(my-function 1)
(my-macro 123)

Racket macro that generates a nested module error

While experimenting with racket's macros, I stumbled into a definition that wasn't at first obvious to me why it was rejected. The code is short and otherwise is probably useless, but is as follows:
#lang racket
(define-syntax (go stx)
(syntax-case stx ()
[(_ id)
#'(module mod racket
(define it id))]
))
(go 'dummy)
The complaint is quote: unbound identifier; also, no #%app syntax transformer...
If I manually inline (define it id) to (define it 'dummy) then it works.
I had a hunch that ' ie. quote of (go 'dummy) that is bound by #lang racket is not recognized as the same binding within the submodule mod even though syntactically it is the same sequence of letters. If I strip 'dummy of all lexical context by round tripping as follows:
(with-syntax ([ok (datum->syntax #f (syntax->datum #'id))])
below the pattern match (_ id) and replace definition of it with (define it ok) then all is good again.
#lang racket
(define-syntax (go stx)
(syntax-case stx ()
[(_ id)
(with-syntax ([ok (datum->syntax #f (syntax->datum #'id))])
#'(module mod racket
(define it ok)))]
))
(go 'dummy)
I presume that my dilemma was caused by the hygiene system. However, is there a more direct solution to convince racket compiler that these identifiers, ie. quote are really the same without this boilerplate?
The expression that you insert for id in:
(module mod racket
(define it id))
is going to be evaluated in the context of the module.
Therefore the syntactic context id id needs to be the same as
the context of the submodule.
You describe one way of removing existing context. Here is another:
#lang racket
(require (for-syntax racket/base))
(define-syntax (go stx)
(syntax-case stx ()
[(_ id)
(with-syntax ([id (syntax->datum #'id)])
#'(module mod racket
(provide it)
(define it id)))]))
(go 42)
(require (submod "." mod))
it
In most macros it is a good thing that context is preserved, so
having to "boiler plate" to remove it seems okay to me.
Of course, if you experience to much boiler plate then write
a macro that inserts the boiler plate for you :-)

Is this Racket macro written in an idiomatic style?

My son and I are learning Racket together and are building a very simple text-based adventure for use directly from the REPL. So, for example, the player can type (go 'north) or (take 'apple).
After getting some basic stuff working, my son thought that quoting the noun was a bit of a pain (strangely, the parens don't bother him!), and so we hacked around with macros for a bit and we did get something working but it required an explicit function and a corresponding macro e.g.
(define (do-take item) ...)
(define-syntax (take stx)
(define item (cadr (syntax->datum stx)))
(datum->syntax stx `(do-take ',item)))
I figured we could do better than this, so I read around a bit more and came up with this:
(require (for-syntax racket/syntax))
(define-syntax (define-verb stx)
(syntax-case stx ()
[(_ (verb noun) body-first body-rest ...)
(with-syntax ([verb-fun (format-id stx "do-~a" #'verb)])
#'(begin
(define-syntax-rule (verb noun) (verb-fun 'noun))
(define (verb-fun noun) body-first body-rest ...)))]))
So now, we can write (define-verb (take item) ...) and the player at the REPL can type (take apple).
My question is whether, given what we want to achieve, this is a reasonable approach or whether there a more simple / idiomatic way to achieve the same thing?
In general, the main thing I would recommend doing is using the syntax/parse library. It has more tools for parsing syntax. You can even use forms like define-syntax-parser to make your macro even more concise. Rewriting your code using syntax/parse (dropping that one line because it doesn't seem to be doing anything), your macro would look like this:
#lang racket
(require syntax/parse/define
(for-syntax syntax/parse racket/syntax))
(define-syntax-parser define-verb
[(_ (verb:id noun) body ...+)
(define/syntax-parse verb-fun (format-id stx "do-~a" #'verb))
#'(begin
(define-simple-macro (verb noun) (verb-fun 'noun))
(define (verb-fun noun) body ...))])
This gives you a few nice things above the example you gave:
the :id ensures that verb is a literal identifier, rather than an expression.
the ...+ means you only need to have one body pattern, rather than two.
Using define/syntax-parse means your code does not get more indented than with-syntax. (Although this one is a matter of preference.)

What is the difference between these macros?

I have some questions about how macros work in Scheme (specifically in Chicken Scheme), let's consider this example:
(define (when-a condition . body)
(eval `(if ,condition
(begin ,#body)
'())))
(define-syntax when-b
(er-macro-transformer
(lambda (exp rename compare)
(let ((condition (cadr exp))
(body (cddr exp)))
`(if ,condition (begin ,#body) '())))))
(define-syntax when-c
(ir-macro-transformer
(lambda (exp inject compare)
(let ((condition (cadr exp))
(body (cddr exp)))
`(if ,condition (begin ,#body) '())))))
(define-syntax when-d
(syntax-rules ()
((when condition body ...)
(if condition (begin body ...) '()))))
Can I consider when-a a macro? I feel that I can't consider it a macro in a strict way since I'm not using define-syntax but I'm not able to say any pratical reason to not prefer this implementation.
Are my macros hygienic?
Is there any difference between when-b and when-c? Since I'm not using rename nor inject I think there isn't.
Can I consider when-a a macro? I feel that I can't consider it a macro in a strict way since I'm not using define-syntax but I'm not able to say any pratical reason to not prefer this implementation.
This works like a macro, but it's not exactly the same as a true macro, for the following reasons:
The main difference between a true macro and your eval-based "macro" is that your approach will evaluate all its arguments before calling it. This is a very important difference. For example: (if #f (error "oops") '()) will evaluate to '() but (when-a #f (error "oops")) will raise an error.
It's not hygienic. Beforehand, one might have done something like (eval '(define if "not a procedure")), for example, and that would mean this eval would fail; the if in the body expression of the "expansion" doesn't refer to the if at the definition site.
It does not get expanded at compile time. This is another major reason to use a macro; the compiler will expand it, and at runtime no computation will be performed to perform the expansion. The macro itself will have completely evaporated. Only the expansion remains.
Are my macros hygienic?
Only when-c and when-d are, because of the guarantees made by ir-macro-transformer and syntax-rules. In when-b you'd have to rename if and begin to make them refer to the versions of if and begin at the macro definition site.
Example:
(let ((if #f))
(when-b #t (print "Yeah, ok")))
== expands to ==>
(let ((if1 #f))
(if1 #t (begin1 (print "Yeah, ok"))))
This will fail, because both versions of if (here annotated with an extra 1 suffix) refer to the same thing, so we'll end up calling #f in operator position.
In contrast,
(let ((if #f))
(when-c #t (print "Yeah, ok")))
== expands to ==>
(let ((if1 #f))
(if2 #t (begin1 (print "Yeah, ok"))))
Which will work as intended. If you want to rewrite when-b to be hygienic, do it like this:
(define-syntax when-b
(er-macro-transformer
(lambda (exp rename compare)
(let ((condition (cadr exp))
(body (cddr exp))
(%if (rename 'if))
(%begin (rename 'begin)))
`(,%if ,condition (,%begin ,#body) '())))))
Note the extra %-prefixed identifiers which refer to the original value of if and begin as they were at the place of definition of the macro.
Is there any difference between when-b and when-c? Since I'm not using rename nor inject I think there isn't.
There is. Implicit renaming macros are called that because they implicitly rename all the identifiers that come in from the usage site, and also every new identifier you introduce in the body. If you inject any identifiers, that undoes this implicit renaming, which makes them unhygienically available for capture by the calling code.
On the other hand, explicit renaming macros are called that because you must explicitly rename any identifiers to prevent them being captured by the calling code.

Generating logging information using Racket

Background:
I intend to generate debug messages for the code I'm developing. I wrote a macro to avoid writing the logging calls in each function. I know this limits the ability to generate more customized debug messages, but in return it isolates logging from code. And that is what I'm aiming for. This macro approach has other drawbacks too, for example it limits creation of function bindings to this macro only, but I think I can live with that.
Following is the definition of the macro and an example demonstrating its usage.
(define-syntax (define-func stx)
(syntax-case stx ()
[(define-func (func-name args ...) body1 body2 ...)
(if (and (identifier? #'func-name)
(andmap symbol? (syntax->datum #'(args ...))))
(syntax (define (func-name args ...)
(log-debug (format "Function-name ~a:" (syntax-e #'func-name)) (list args ...))
body1
body2 ...))
(raise-syntax-error 'define-func "not an identifier" stx))]
[else (raise-syntax-error 'define-func "bad syntax" stx)]))
(define-func (last l)
(cond [(null? l) null]
[(null? (rest l)) (first l)]
[else (last (rest l))]))
(define-func (main)
(last (list 1 2 3 4 5 6 7 8 9))
(logger))
log-debug and logger are defined in separate module
The output produced is somewhat like following:
Function-name last:
args:
:-> (7 8 9)
Function-name last:
args:
:-> (8 9)
Function-name last:
args:
:-> (9)
Now I want to make it more readable. By readability I mean to provide some kind of indentation so that the person reading the log could make sense of call flow. For example something like following:
Function-name last:
args:
:-> (7 8 9)
Function-name last:
args:
:-> (8 9)
Function-name last:
args:
:-> (9)
It is easier to figure out who called whom and so forth. I have an idea that can do this. It involves a variable that keeps track of indentation then after logging the function name I will increase the indent and after evaluation of body and before returning the value decrements the value. Something like following:
(define indent 0)
(define-syntax (define-func stx)
(syntax-case stx ()
[ (... ...)
(...
(log-debug ...)
(increment indent)
(let [(retval (body1 body2 ...)]
(decrease indent)
retval))]))
increment and decrease increases and decreases indentation respectively.
Problem:
It works even for function that returns void. I'm not sure whether its the correct behavior. In racket void is a special value, but I'm not sure that creating a binding to void is right way to go.
Is there any better way to achieve the same? If not are there any problems in this design? I'm open to any idea/change as long as they keep the logging and code separate.
thanks for the help!
I have several suggestions for you:
It's probably better to use a parameter instead of a variable, for "global" stuff like your indentation level, since the original value is restored for you at the end of the parameterize expression.
All those raise-syntax-error checks you have in your macro are totally superfluous: syntax-case already provides guards (also known as fenders) that allow you to do any validation of macro "arguments" necessary:
(define-syntax (define-func stx)
(syntax-case stx ()
[(_ (func-name args ...) body1 body2 ...)
(andmap identifier? (syntax->list #'(func-name args ...)))
#'(define (func-name args ...)
(log-debug (format "Function-name ~a:" 'func-name)
(list args ...))
body1
body2 ...)]))
I've also fixed up your code in several places, as you can see above:
I used (_ ...) instead of (define-func ...), since in syntax-case (unlike syntax-rules), the latter will actually bind a pattern variable called define-func, which will affect any recursive macro calls you may want to do (I'll grant that you don't have one here, but it's a good habit to have anyway).
Rather than completely flatten the #'(args ...) in the guard, I just turned it into a list of syntax objects so you can test using identifier?. This is more intention-revealing than testing using symbol?, and allows us to also test func-name in the same expression.
You don't need to use (syntax-e #'func-name) inside the expanded code! Just quote it.