I'm running up against a problem in understanding the CLOS way of handling file access within a class. In c++ I would be able to do this:
class Foo {
Foo (string filename); // opens the file (my_file) requested by the filename
~Foo (); // close the file
FILE * my_file; // a persistent file-handle
DataStruct my_data; // some data
void ParseData (); // will perform some function on the file and populate my_data
DataStruct * GetData () { return &my_data; } // accessor to the data
};
What I'd like to point out is that PraseData() will be called multiple times, and each time a new block of data will be parsed from the file and my_data will be altered.
I'm trying to perform the same trick in CLOS - create all the generic methods to parse the data, load the file, read headers, etc. as well as the class definition which I have as:
(defclass data-file ()
((filename :initarg :filename :accessor filename)
(file :accessor file)
(frame :accessor frame)))
In the "constructor" (i.e. initialize-instance) I open the file just as my c++ idiom. Then I have access to the data and I can parse the data as before. However, I'm told that using a "destructor" or (finalize) method to close the file is not idiomatic CLOS for handling this type of situation where I need the file to be around so I can access it outside of my data-file methods.
I'm going to define a function that loads a data-file, and then performs a series of analyses with its data, and then hopefully close it. What's a way to go about doing this? (I'm assuming a macro or some type of closure would work in here, but I'm not familiar enough with the lisp way to decide what is needed or how to implement it).
One option is to have the stream as a slot instead of the filename, and then scope it with WITH-OPEN-FILE:
(with-open-file (stream file)
(let ((foo (make-instance 'foo :stream stream)))
(frob foo)
(...other processing of foo...)))
Then your stream will be closed automatically.
I think I would lean towards making classes only to store complete authoritative data (what you call DataStruct?).
You don't really need a special class for "loading + storage of another class". Plus, that way has the unspoken invariant that my_data holds the data of my_file up to the current seek position, which seems a bit strange to my eye.
Put another way: what does Foo do? Given a filename, it loads data, and gives you a DataStruct. That sounds like a function to me. If you need to be able to run it in a thread, or fire events between loading records, a class is the natural way to do it in C++, but you don't need a class for those things in Lisp.
Also, remember that you don't need to use DEFCLASS in order to use generic methods in CLOS.
I don't know what the structure of your data is, but in similar situations I've made a parse-one-chunk function that takes a stream and returns one record, and then create a complete Foo instance inside a loop in a with-open-file. If the stream is never needed outside the scope of a with-open-file expansion, you never need to worry about closing it.
Related
This code:
(require racket/generic)
;; A holder that assigns ids to the things it holds. Some callers want to know the
;; the id that was assigned when adding a thing to the holder, and others don't.
(define-generics holder
(add-new-thing+id holder new-thing) ;returns: holder id (two values)
(add-new-thing holder new-thing) ;returns: holder
#:fallbacks
[(define (add-new-thing holder new-thing) ;probably same code for all holder structs
(let-values ([(holder _) (add-new-thing+id holder new-thing)])
holder))])
produces this error message:
add-new-thing+id: method not implemented in: (add-new-thing+id holder new-thing)
I'm able to fix it by adding a define/generic inside the fallbacks, like this:
#:fallbacks
[(define/generic add add-new-thing+id)
(define (add-new-thing holder new-thing)
(let-values ([(holder _) (add holder new-thing)])
holder))])
but this seems to add complexity without adding value, and I don't understand why one works and the other doesn't.
As I understand #:fallbacks, the idea is that the generic definition can build methods out of the most primitive methods, so structs that implement the generic interface don't always need to reimplement the same big set of methods that usually just call the core methods with identical code—but they can override those "derived" methods if needed. say, for optimization. That's a very useful thing*—but have I misunderstood fallbacks?
It seems strange that fallbacks code couldn't refer to the generic methods. Isn't the main value of fallbacks to call them? The documentation for define/generic says that it's a syntax error to invoke it outside a #:methods clause in a struct definition, so I'm probably misusing it. Anyway, can someone explain what are the rules for code in a #:fallbacks clause? How are you supposed to write it?
* The Clojure world has something similar, in the potemkin library's def-abstract-type and deftype+, but not as well integrated into the language. potemkin/def-map-type illustrates very nicely why fallbacks—as I understand them, anyway—are such a valuable feature.
The second version of your code is correct.
The first version of your code would work if you had a fallback definition of add-new-thing+id, but because you are referring to any possible definition of that method outside of the fallback scope, you need to import it.
It effectively feels a bit repetitive to have to define the generic again inside the fallback clause. It's because #:fallbacks works the same way as #:methods, and therefore has the same behavior of overriding generics with its own definitions.
To make it explicit that you are overriding a method, you need to "import" it inside your clause, using define/generic (which is not really defining anything, it is just importing the generic into the context).
As the documentation for define/generic says:
When used inside the method definitions associated with the #:methods keyword, binds local-id to the generic for method-id. This form is useful for method specializations to use generic methods (as opposed to the local specialization) on other values.
Then in define-generics:
The syntax of the fallback-impls is the same as the methods provided for the #:methods keyword for struct.
Which means #:fallbacks has the same behavior as using #:methods in a struct.
Why?
The logic behind that behavior is that method definition blocks, like #:methods and #:fallbacks have access to their own definitions of all the generics, so that it's easy to refer to your own context. To explicitly use a generic from outside this context, you need to use define/generic.
I want to make a macro, that when used in class definition creates a field, it's public setter, and an annotation. However, it'd seem the macro is not expanding, mostly because it's used inside other (class definition) macro.
Here is an example how to define a class with one field:
(define-simple-class test-class ()
(foo :: java.util.List ))
My macro (only defines field as of now):
(define-syntax autowire
(syntax-rules ()
((autowire class id)
(id :: class))))
However, if I try to use it:
(define-simple-class test-class ()
(autowire java.util.List foo))
and query fields of the new class via reflection, I can see that it creates a field named autowire, and foo is nowhere to be seen. Looks like an issue of the order the macros are expanded.
Yes, macros are expanded “from the outside in”. After expanding define-simple-class, the subform (autowire java.util.List foo) does not exist anymore.
If you want this kind of behaviour modification, you need to define your own define-not-so-simple-class macro, which might expand to a define-simple-class form.
However, please step back before making such a minor tweak to something that is standard, and ask yourself whether it is worth it. The upside might be syntax that is slightly better aligned to the way you think, but the downside is that it might be worse aligned to the way others think (who might need to understand your code). There is a maxim for maintainable and readable coding: “be conventional”.
I've run into a problem that a third-party library needs to act on a class as if it was finalized. After some reading I understand the motivation behind this mechanism, but I don't really know how it functions.
Example:
(make-instance 'expression :op '+ :left 'nan :right 'nan)
(defmethod normalize-expression ((this expression))
(optima:match this
((optima::or (expression :left 'nan) (expression :right 'nan)) 'nan)
((expression :op op :left x :right y) (funcall op x y))))
Unless I add the first line, the function will not compile, giving me this error:
; caught ERROR:
; (during macroexpansion of (SB-PCL::%DEFMETHOD-EXPANDER NORMALIZE-EXPRESSION ...))
; SB-MOP:CLASS-SLOTS called on #<STANDARD-CLASS EXPRESSION>, which is not yet finalized.
; See also:
; AMOP, Generic Function SB-MOP:CLASS-SLOTS
optima is a pattern-matching library, the (expression :op op ...) is matching instances of class expression against the given pattern. I don't know in much details, but it looks like it needs to know what are the accessors defined for this class, and it looks like that information is not available until it is finalized. So, is there any way to sidestep the finalization problem?
The class will not be extended (at least not in this project, and it's not being planned). It doesn't hurt that much to create a dummy instance... it is just an ugly solution, so I hoped to find a better one. Also, perhaps, I'd get some more info on finalization, which is good too :)
Forgetting to ensure class finalization seems to be quite common mistake when using MOP.
In lisp, classes are defined in two "phases":
Direct class definition
Effective class definition
Direct class definition is isomorphic to defclass form. It has class name, names of superclasses, list of direct slots (i.e., slots defined on this particular class but on its superclasses).
Effective class definition contains all information needed for compiler/interpreter. It contains list of all class slots (including those defined on superclasses), class instance layout, references to accessor methods, etc.
Process of transforming direct class definition to effective class definition is called class finalization. Since CLOS supports redefining classes, finalization might be called multiple times for a class. One of the reasons why finalization is delayed is because class may be defined before its superclasses are defined.
Regarding your particular problem: is seems that optima:match should ensure that class is finalized before trying to list its slots. This can be done with two functions: class-finalized-p (to check whether class needs finalization) and finalize-inheritance to actually perform finalization. Or you can use utility function closer-mop:ensure-finalized. (closer-mop is a library for portable usage of CLOS MOP).
E.g.,:
(c2mop:ensure-finalized (find-class 'expression))
I'm trying to implement the ideas from http://thinkrelevance.com/blog/2013/06/04/clojure-workflow-reloaded into my codebase.
I have a dao layer, where I now need to pass in a database in order to avoid global state. One thing that is throwing me off is the phrase:
Any function which needs one of these components has to take it as a
parameter. This isn't as burdensome as it might seem: each function
gets, at most, one extra argument providing the "context" in which it
operates. That context could be the entire system object, but more
often will be some subset. With judicious use of lexical closures, the
extra arguments disappear from most code.
Where should I use closures in order to avoid passing global state for every call? One example would be to create an init function in the dao layer, something like this:
(defprotocol Persistable
(collection-name [this]))
(def save nil)
(defn init [{:keys [db]}]
(alter-var-root #'save (fn [_] (fn [obj] (mc/insert-and-return db (collection-name obj) obj WriteConcern/SAFE)))))
This way I can initiate my dao layer from the system/start function like this:
(defn start
[{:keys [db] :as system}]
(let [d (-> db
(mc/connect)
(mc/get-db "my-test"))]
(dao/init d)
(assoc system :db d)))
This works, but it feels a bit icky. Is there a better way? If possible I would like to avoid forcing clients of my dao layer to have to pass a database every time it uses a function.
You can use higher order function to represent your DAO layer - that's the crux of functional programming, using functions to represent small to large parts of your system. So you have a higher order function which takes in the DB connection as param and return you another function which you can use to call various operations like save, delete etc on the database. Below is one such example:
(defn db-layer [db-connection]
(let [db-operations {:save (fn [obj] (save db-connection obj))
:delete (fn [obj] (delete db-connection obj))
:query (fn [query] (query db-connection query))}]
(fn [operation & params]
(-> (db-operations operation) (apply params)))))
Usage of DB layer:
(let [my-db (create-database)
db-layer-fn (db-layer my-db)]
(db-layer-fn :save "abc")
(db-layer-fn :delete "abc"))
This is just an example of how higher order functions can allow you to sort of create a context for another set of functions. You can take this concept even further by combining it with other Clojure features like protocols.
In common-lisp, I want to implement a kind of reference system like this:
Suppose that I have:
(defclass reference () ((host) (port) (file)))
and also I have:
(defun fetch-remote-value (reference) ...) which fetches and deserializes a lisp object.
How could I intervene in the evaluation process so as whenever a reference object is being evaluated, the remote value gets fetched and re-evaluated again to produce the final result?
EDIT:
A more elaborate description of what I want to accomplish:
Using cl-store I serialize lisp objects and send them to a remote file(or db or anything) to be saved. Upon successful storage I keep the host,port and file in a reference object. I would like, whenever eval gets called on a reference object, to first retrieve the object, and then call eval on the retrieved value. Since a reference can be also serialized in other (parent) objects or aggregate types, I can get free recursive remote reference resolution by modyfing eval so i dont have to traverse and resolve the loaded object's child references myself.
EDIT:
Since objects always evaluate to themselves, my question is a bit wrongly posed. Essentially what I would like to do is:
I would like intercept the evaluation of symbols so that when their value is an object of type REFERENCE then instead of returning the object as the result of the symbol evaluation, to return the result of (fetch-remote-value object) ?
In short: you cannot do this, except by rewriting the function eval and modifying your Lisp's compiler. The rules of evaluation are fixed Lisp standard.
Edit After reading the augmented question, I don't think, that you can achieve full transperency for your references here. In a scenario like
(defclass foo () (reference :accessor ref))
(ref some-foo)
The result of the call to ref is simply a value; it will not be considered for evaluation regardless of its type.
Of course, you could define your accessors in a way, which does the resolution transparently:
(defmacro defresolver (name class slot)
`(defmethod ,name ((inst ,class))
(fetch-remote-reference (slot-value inst ',slot))))
(defresolver foo-reference foo reference)
Edit You can (sort of) hook into the symbol resolution mechanism of Common Lisp using symbol macros:
(defmacro let-with-resolution (bindings &body body)
`(symbol-macrolet ,(mapcar #'(lambda (form) (list (car form) `(fetch-aux ,(cadr form)))) bindings) ,#body))
(defmethod fetch-aux ((any t)) any)
(defmethod fetch-aux ((any reference)) (fetch-remote-reference any))
However, now things become pretty arcane; and the variables are no longer variables, but magic symbols, which merely look like variables. For example, modifying the content of a variable "bound" by this macro is not possible. The best you can do with this approach is to provide a setf expansion for fetch-aux, which modifies the original place.
Although libraries for lazy evaluatione and object persistence bring you part of the way, Common Lisp does not provide a portable way to implement fully transparent persistent values. Lazy or persistent values still have to be explicitly forced.
MOP can be used to implement lazy or persistent objects though, with the slot values transparently forced. It would take a change in the internals of the Common Lisp implementations to provide general transparency, so you could do e.g. (+ p 5) with p potentially holding a persistent or lazy value.
It is not possible to directly change the evaluation mechanisms. You would need to write a compiler for your code to something else. Kind of an embedded language.
On the CLOS level there are several ways to deal with it:
Two examples:
write functions that dispatch on the reference object:
(defmethod move ((object reference) position)
(move (dereference reference) position))
(defmethod move ((object automobile) position)
...))
This gets ugly and might be automated with a macro.
CHANGE-CLASS
CLOS objects already have an indirection, because they can change their class. Even though they may change their class, they keep their identity. CHANGE-CLASS is destructively modifying the instance.
So that would make it possible to pass around reference objects and at some point load the data, change the reference object to some other class and set the slots accordingly. This changing the class needs to be triggered somewhere in the code.
One way to have it automagically triggered might be an error handler that catches some kinds of errors involving reference object.
I would add a layer on top of your deserialize mechanism that dispatches based on the type of the incoming data.