I'm trying to add completion at point for frame-local variables from backtrace-frames during invocations of read--expression by debugger-eval-expression or edebug-eval-expression.
I constructed the following completion table to add frame-local variables to the already available table for local elisp variables,
;; completion table for locals in current frame
(defvar my-backtrace-locals-completion-table
(completion-table-in-turn
(completion-table-dynamic
(lambda (_string)
(when-let* ((idx (backtrace-get-index)) ;backtrace.el
(frame (nth idx backtrace-frames)))
(backtrace-frame-locals frame)))
'do-switch-buffer)
elisp--local-variables-completion-table)) ;elisp-mode.el
which seems to work fine, eg. to reproduce
(1) evaluate
;; debug-on-error = t
(let ((my-local-var '(1 2))) (mapcan #'car this-local-var))
(2) from debugger's second frame, evaluate with eval-expression
(funcall my-backtrace-locals-completion-table "my-" nil t)
returns expected ("my-local-var").
The problem is following the above steps, but calling instead calling debugger-eval-expression doesn't work. The environment where the table is evaluated isn't finding a backtrace-frame (with or without do-switch-buffer).
How can I define the table to be evaluated in the proper buffer?
emacs v27.0.50
The completion table above doesn't quite return the expected candidates for debugger-eval-expression. The environment where the expression is evaluated has locals from frames higher than, but not including, the one at point in the Backtrace buffer.
So, the available locals should be only those from higher frames, eg.
(eval-when-compile (require 'dash))
(defvar my-backtrace-locals-completion-table
(completion-table-dynamic
(lambda (_string)
(when backtrace-frames
(--mapcat
(-some->> (backtrace-frame-locals it) (--map (car it)))
(nthcdr (1+ (backtrace-get-index)) backtrace-frames))))
'do-switch-buffer))
Then, redefining debugger-eval-expression's interactive spec to use the new locals table in place of the normal elisp table provides the correct completions (passing the 'do-switch-buffer arg completion-table-dynamic to find completions in the original buffer).
(defun my-backtrace#eval (orig-fn exp &optional nframe)
(interactive
(let ((elisp--local-variables-completion-table
my-backtrace-locals-completion-table))
(list (read--expression "[my] Eval in stack frame: "))))
(apply orig-fn (list exp nframe)))
(advice-add 'debugger-eval-expression :around #'my-backtrace#eval)
Related
Update: The original version of this question did not fully describe the constraints of my situation. It now includes an example with some hand-waving for the sake of simplicity as well as a testable minimum example.
I am trying to pass a snippet of code into a procedure for evaluation at a later point in time. The system with all of the constraints looks like this:
; External Library code - cannot be changed
(define (serialize-parameters . args)
(format some-file "~A~%" args))
(define (function-called-later-possibly-after-a-system-reboot callback)
(apply callback some-default-parameters (deserialize-parameters some-file)))
; Internal library code - can be changed, but the value of `code-snippet` cannot be hardcoded
(define (callback some-default-parameters code-snippet)
(eval code-snippet (current-module))
; Application code - can be changed, can be hardcoded
(define (main)
(serialize-parameters #:code-snippet '(format #t "~A~%" some-default-parameters)))
The main point is to allow the application to execute an arbitrary code snippet in a callback function. The callback function receives other parameters which are not known to the application and may change between calls.
An important constraint is that the parameters that the application wants to send to the callback are written to a file and then read back at a later point in time, possibly after a system reboot.
This example is easier to test, but does not capture all of the constraints:
(define (test foo bar)
(eval bar (current-module)))
(test "value of foo" '(format #t "~A~%" foo))
Running this program results in Unbound variable: foo. What I want is for the test function to be defined in such a way that the result of the call will be "value of foo\n" being printed to the terminal. Is this possible in Guile?
Thank you.
This will no work. eval takes an environment and you pass it (current-module). That are top level bindings in the module, like test but not any lexical bindings like foo or bar. They just don't exist in the environment returned by (current-module).
You can do this:
(define foo "value of foo")
(eval '(format #t "~A~%" foo) (current-module))
; prints "value of foo"
; ==> #t
Also, the elephant in the room is that you can do this with lambdas:
(define (test value proc)
(proc value))
(test "value to process" (lambda (foo) (format #t "~A~%" foo)))
; prints "value to process"
; ==> #t
Alternatively, but I'm guessing you can't have format in callback because "code-snippet" can have many different values:
(define (main)
(serialize-parameters #:code-snippet "~A~%"))
(define (callback some-default-parameters code-snippet)
(format #t code-snippet some-default-parameters))
EDIT
I think you can do it semi hard-coded:
(define (main)
(serialize-parameters #:code-snippet 'print-format))
(define (callback argument message)
(case message
((print-format) (format #t "~A~%" argument))
((other-message) (handle-message ...))
(else ...)))
You can even make it a dynamic dispatcher. Eg. you do something like this:
(define messages '())
(define (register-callback message proc)
(set! messages (cons (cons message proc) messages)))
(define (callback argument message)
(let ((found (assq message messages)))
(when found
((cdr found) argument))))
(register-callback 'print-format (lambda (arg) (format #t "~A~%" arg)))
(callback "test" 'print-format) ; prints "test"
Now only the message gets stored in a file, which easily can be any data literal.
I am currently reading the chapter on read-time macros from Paul Graham's "On Lisp" book.
The problem I am encountering is the following. When I run one of his examples:
(set-dispatch-macro-character #\# #\?
#’(lambda (stream char1 char2)
‘#’(lambda (&rest ,(gensym))
,(read stream t nil t))))
I get the following error:
No dispatch function defined for #\’
Why is it happening? Could it be because I am running it at the REPL? What can one do to fix it?
The PDF from which you are copying the code uses punctuation marks outside the range of basic ASCII characters you are supposed to use here:
CL-USER> (char-name #\’)
"RIGHT_SINGLE_QUOTATION_MARK"
The usual quote symbol should instead use the apostrophe character:
CL-USER> (char-name #\')
"APOSTROPHE"
The same goes for backquote:
CL-USER> (char-name #\‘)
"LEFT_SINGLE_QUOTATION_MARK"
You should be writing instead:
(set-dispatch-macro-character #\# #\?
#'(lambda (stream char1 char2)
`#'(lambda (&rest ,(gensym))
,(read stream t nil t))))
The #' is not necessary before lambda, since Common Lisp also defines a macro named lambda which expands into (function (lambda ...)).
You can test your new read macro as follows:
CL-USER> #?10
#<FUNCTION (LAMBDA (&REST #:G617)) {1001C541FB}>
CL-USER> (funcall *)
10
When using SBCL, I obtain warnings about unused variables. This happens because the code declares variables in anonymous functions but never uses them. This is not a serious problem, but generally speaking, it is better to declare which variables are ignored:
(set-dispatch-macro-character
#\# #\?
(lambda (stream &rest chars)
(declare (ignore chars))
(let ((rest (gensym)))
`(lambda (&rest ,rest)
(declare (ignore ,rest))
,(read stream t nil t)))))
I will describe my problem on example.
I'll get (play '(left nothing right left)). Some of the names in the list are real procedures, others i want to skip.
(define (left)
'left
)
I need to interpret procedures with names in the list. What is the solution?
When I try ( (car '(left nothing right left))) I get error : procedure application: expected procedure, given: left (no arguments)
(car '(left nothing right left)) evaluates to the symbol left, which is the name of a procedure, but not actually a procedure, so you can't call it.
You'll want to build an association list mapping symbols to procedures:
(define actions `((left . ,(lambda () 'left))
(right . ,(lambda () 'right))
(nothing . ,(lambda () (display "I'm staying put")))))
then you can call the appropriate function for the first element in your list as
((cdr (assoc (car '(left nothing right left)) actions)))
You can also use quasiquoting to construct a list containing a mixture of symbols you want evaluated and others you don't, e.g.
(play `(,left nothing nothing ,right nothing))
left and right will expand to whatever you've defined them as (such as a procedure) while nothing is not un-quoted so it will be left as a symbol. play would then have to test each member to see if it's a procedure, something like:
(define (play xs)(for-each (lambda (x)(if (procedure? x)(x) x)) xs))
(define bootstrap-c-code
(define (from-file file-name)
(let* ((ip (open-input-file file-name))
(res (read-text-file-from-input-port ip)))
(close-input-port ip)
res))
(from-file "llvm.c"))
Error : define: bad syntax (multiple expressions after identifier)
But I can't see anything wrong with it. Can someone explain / fix it please.
It's not clear what you intended with the above code. If you were trying to load a text file and leave the loaded value in a variable called bootstrap-c-code, then try this:
(define bootstrap-c-code
(let ((from-file
(lambda (file-name)
(let* ((ip (open-input-file file-name))
(res (read-text-file-from-input-port ip)))
(close-input-port ip)
res))))
(from-file "llvm.c")))
Of course, the from-file definition will only be visible inside the let, if you need to use it outside, define it outside of the whole expression. If you only need the functionality of from-file inside the let, you can obtain the same result in a much simpler way:
(define bootstrap-c-code
(let* ((ip (open-input-file "llvm.c"))
(res (read-text-file-from-input-port ip)))
(close-input-port ip)
res))
On the other hand, if what you intended was to create a procedure called bootstrap-c-code, then the correct syntax would be:
(define (bootstrap-c-code)
(define (from-file file-name)
(let* ((ip (open-input-file file-name))
(res (read-text-file-from-input-port ip)))
(close-input-port ip)
res))
(from-file "llvm.c"))
According to R5RS, internal definitions can occur only at the beginning of the of a bunch of forms like let, let*, lambda etc. In the case of your code, that is not the case since you have an internal definition inside a non-procedural define. You could fix it by making `bootstrap-c-code' bind to a procedure.
I'd like to export or replicate a scheme environment in another guile process. The algorithm I'm imagining would do something like this to serialize:
(map (lambda (var val) (display (quasiquote (define ,var ,val))
(newline))
(get-current-environment))
And then I'd read/eval that on the other end.
However, while there are functions that return the current environment, they are in some internal format that I can't just map across. How can I "walk" the environment as the above? Alternatively, how else can I replicate an environment into another process?
you may decompose the so-called "current-environment" like this:
(define (get-current-binding-list)
(let* ((e (current-module)) ;; assume checking current-module
(h (struct-ref e 0)) ;; index 0 is current vars hashtable
)
(hash-map->list cons h) ;; return a vars binding list
))
and you can call (get-current-binding-list) to get variables binding list in current-module.
Please note that each element in this list is a pair of symbol and variable type, say, (symbol-name . variable-type). So you may print it like this:
for a instance ,you got a var binding:
(define abc 5)
then:
(let ((vl (get-current-binding-list)))
(assoc-ref vl 'abc)
)
==> #<variable 9bb5108 value: 5>
This result is a "variable type" of variable "abc". You can get it's value with variable-ref procedure.
So you can trace all the bindings and do something ,in your code ,it's simply print var-name and var-value.
I know my answer is too brief, but I think there's enough information to help you to find more details in the manual.
Hope this will help you.
You can't really serialize Scheme environment. I don't known even it's possible to (portably) serialize continuations. Oh, and don't forget about FFIs. Ports and threads are unserializable too.