Using core.matrix in a ClojureScript project. For javascript interop I need aljabr as explained in https://github.com/mikera/core.matrix/wiki/Matrix-implementations
After I try to specify the implementation (as I do with other Clojure implementations)
(require '[clojure.core.matrix :as mat])
(require '[thinktopic.aljabr.core :as imp])
(mat/set-current-implementation :aljabr)
I get this error and I can't use the matrix implementation for Javascript
INFO: No dynamic loading of implementations in Clojurescript.
You must require an implementation explicitly in a namespace, for example thinktopic.aljabr.core
Looking on github I only find the implementation used as I mention above (e.g. https://github.com/mars0i/free/commit/71dbbe4d58645ad4e25f2ac2d4ccba6ccef93968), how can I make aljabr work in cljs?
You need to set the implementation statically, that means not at the beginning of the file like you did but when you actually declare your matrix
(mat/matrix :aljabr [[1.0 2.0 3.0][4.0 5.0 7.0]])
(I answered to myself just cause there is no result on the internet for that error and took me a while to understand the error message explanation, even with git blames)
Related
I have a project with some NPM deps, noteably "#grafana/data": "7.0.1". These dependencies bring in several other libs like apache-arrow. When I try to require deps via (require '"#grafana/data"), I get a bunch of compiler errors like:
ERROR: JSC_LANGUAGE_FEATURE. This language feature is only supported for ECMASCRIPT8 mode or better: async function.
I found similar kinds of errors addressed in google closure compiler issues which leads me to beleive that I cant have library dependencies that require modern javascript as the closure compiler can't handle them. Similarly the closurescript docs don't allow for a :language-in argument for es-2018 et. al.
The the clojure build.edn settings I set are:
:language-in :es-next
:language-out :es5
:rewrite-polyfills
Unfortunately, this doesn't work.
Is there any way that I can get the closure compiler to handle modern javascript dependencies? Are there some crazy hacks that I can do to make this work?
In my ClojureScript code I am requiring a JavaScript module called seedrandom which is in the node_modules folder, like this:
(ns something.core
(:require ["seedrandom" :as rnd]))
(js/console.log (.quick (rnd "x")))
According to the seedrandom documentation it is intended for both nodejs and the browser, and I've previously included and used it successfully in ClojureScript code via a <script> tag, confirming it works in the browser.
Running this cljs file in lumo on the command line works well and outputs a deterministically random number.
When I try to use this same cljs file in my Reagent frontend project I see the following error:
Compiling build :app to "public/js/app.js" from ["src" "env/dev/cljs"]...
events.js:183
throw er; // Unhandled 'error' event
^
Error: module not found: "crypto" from file /home/chrism/dev/something/node_modules/seedrandom/seedrandom.js
at onresolve (/home/chrism/dev/something/node_modules/#cljs-oss/module-deps/index.js:181:30)
...
Inside seedrandom.js we see the following:
// When in node.js, try using crypto package for autoseeding.
try {
nodecrypto = require('crypto');
} catch (ex) {}
Clearly this code is intended to ignore the built-in nodejs crypto module when running in the browser. The problem, as far as I can tell, is that the ClojureScript compiler does not know that - it sees that require('crypto') and tries to pull it into the compilation phase, but can't find it because it's a nodejs built-in.
Is there some way I can tell the compiler to ignore that particular require? Or can I shim the 'crypto' module somehow? What is the cleanest way to solve this?
Note: I have previously experienced this same issue with JavaScript modules which check for the fs node module. Hope we can find a general solution to use again in future. Thanks!
Relevant versions: [org.clojure/clojurescript "1.10.520"] and [reagent "0.8.1"].
This answer is related, asking a similar question from the perspective of Google Closure, which ClojureScript uses, but I'm looking for an answer I can use specifically with cljs.
I have a library of functions which I want to let users play with in the browser.
So I want to set up a situation like this :
I'm developing with figwheel and devcards.
In the main core.cljs I require various functions from my library, so they're all in scope.
Now I want to let the user enter some code which calls that library.
I see how I can run that code with eval, but I can't see how to make my library functions visible to the code being evaled.
And I'm confused by most of the documentation I'm seeing about this (eg. How can I make functions available to ClojureScript's eval?)
Is it possible? And if so, does anyone have a simple example of it being done?
cheers
Phil
Yes, it is possible to provide access to an ambient / pre-compiled library used by evaluated code.
First, you must ensure that the functions in your library are available in the JavaScript runtime. In other words, avoid :advanced optimization, as this will eliminate functions not called at compile time (DCE). Self-hosted ClojureScript is compatible with :simple.
Second, you need to make the analysis metadata available to the self-hosted compiler that will be running in the browser (either making use of cljs.js/load-analysis-cache! or an optional argument to cljs.js/empty-state).
A minimal project illustrating how to do this is below (and also at https://github.com/mfikes/ambient):
Project Code
src/main/core.cljs:
(ns main.core
(:require-macros [main.core :refer [analyzer-state]])
(:require [cljs.js]
[library.core]))
(def state (cljs.js/empty-state))
(defn evaluate [source cb]
(cljs.js/eval-str state source nil {:eval cljs.js/js-eval :context :expr} cb))
(defn load-library-analysis-cache! []
(cljs.js/load-analysis-cache! state 'library.core (analyzer-state 'library.core))
nil)
src/main.core.clj:
(ns main.core
(:require [cljs.env :as env]))
(defmacro analyzer-state [[_ ns-sym]]
`'~(get-in #env/*compiler* [:cljs.analyzer/namespaces ns-sym]))
src/library/core.cljs:
(ns library.core)
(defn my-inc [x]
(inc x))
Usage
We have a main.core namespace which provides an evaluate function, and this example will show how to call functions in an ambient / pre-compiled library.core namespace.
First, start up a browser REPL via
clj -m cljs.main
At the REPL, load our main namespace by evaluating
(require 'main.core)
Test that we can evaluate some code:
(main.core/evaluate "(+ 2 3)" prn)
This should print
{:ns cljs.user, :value 5}
Now, since main.core required library.core, we can call functions in that namespace. Evaluating this at the REPL will yield 11:
(library.core/my-inc 10)
Now, let's try to use this "ambient" function from self-hosted ClojureScript:
(main.core/evaluate "(library.core/my-inc 10)" prn)
You will see the following
WARNING: No such namespace: library.core, could not locate library/core.cljs, library/core.cljc, or JavaScript source providing "library.core" at line 1
WARNING: Use of undeclared Var library.core/my-inc at line 1
{:ns cljs.user, :value 11}
In short, what is going on is that even though library.core.my_inc is available in the JavaScript environment, and can indeed be called, producing the correct answer, you get warnings from the self-hosted compiler that it knows nothing about this namespace.
This is because the compiler analysis metadata is not in the main.core/state atom. (The self-hosted compiler has its own analysis state, held in that atom in the JavaScript environment, which is separate from the JVM compiler analysis state, held via Clojure in the Java environment.)
Note: If we instead had the source for library.core compiled by the self-hosted compiler (by perhaps by using main.core/evaluate to eval "(require 'library.core)", along with properly defining a cljs.js/*load-fn* that could retrieve this source, things would be good, and the compiler analysis metadata would be in main.core/state. But this example is about calling ambient / pre-compiled functions in library.core.
We can fix this by making use of cljs.js/load-analysis-cache! to load the analysis cache associated with the library.core namespace.
This example code embeds this analysis cache directly in the code by employing a macro that snatches the analysis cache from the JVM-based compiler. You can transport this analysis cache to the browser by any mechanism you desire; this just illustrates one way of simply embedding it directly in the shipping code (it's just data).
Go ahead and evaluate the following, just to see what the analysis cache for that namespace looks like:
(main.core/analyzer-state 'library.core)
If you call
(main.core/load-library-analysis-cache!)
this analysis cache will be loaded for use by the self-hosted compiler.
Now if you evaluate
(main.core/evaluate "(library.core/my-inc 10)" prn)
you won't see any warnings and this will be printed:
{:ns cljs.user, :value 11}
Furthermore, since the self-hosted compiler now has the analysis metadata for libraray.core, it can properly warn on arity errors, for example
(main.core/evaluate "(library.core/my-inc 10 12)" prn)
will cause this to be printed:
WARNING: Wrong number of args (2) passed to library.core/my-inc at line 1
The above illustrates what happens when you don't have the analyzer cache present for a namespace and how to fix it using cljs.js/load-analysis-cache!. If you know that you will always want to load the cache upon startup, you can simply things, making use of an optional argument to cljs.js/empty-state to load this cache at initialization time:
(defn init-state [state]
(assoc-in state [:cljs.analyzer/namespaces 'library.core]
(analyzer-state 'library.core)))
(def state (cljs.js/empty-state init-state))
Other Projects
A few (more elaborate) projects that make library functions available to self-hosted ClojureScript in the browser:
Klangmeister
power-turtle
life-demo
I have a Clojurescript program running in the browser.
It imports a number of libraries, and then I want to allow the user to enter some small clojurescript "glue-code" that calls those libraries.
I can see (from https://cljs.github.io/api/cljs.js/eval) that you call eval with four arguments, the first being the state of the environment, which is an atom. But can I actually turn my current environment with all the functions I've required from elsewhere, into an appropriate argument to eval?
Update :
I thought that maybe I could set the namesspace for the eval using the :ns option of the third, opts-map, argument. I set it to the namespace of my application :
:ns "fig-pat.core"
But no difference.
Looking at the console, it's definitely the case that it's trying to do the evaluation, but it's complaining that names referenced in the eval-ed code are NOT recognised :
WARNING: Use of undeclared Var /square
for example. (square is a function I'm requiring. It's visible in the application itself ie. the fig-pat.core namespace)
I then get :
SyntaxError: expected expression, got '.'[Learn More]
Which I'm assuming this the failure of eval-ed expression as a whole.
Update 2 :
I'm guessing this problem might actually be related to : How can I get the Clojurescript namespace I am in from within a clojurescript program?
(println *ns*)
is just printing nil. So maybe Clojurescript can't see its own namespace.
And therefore the :ns in eval doesn't work?
Calling eval inside a clojurescript program is part of what is called "self-hosted clojurescript".
In self-hosted clojurescript, namespaces are not available unless you implement a resolve policy. It means that have to let the browser know how to resolve the namespace e.g. loads a cljs file from a cdn.
It's not so trivial to implement namespace resolving properly.
This is explained in a cryptic way in the docstring of load-fn from cljs.js namespace.
Several tools support namespaces resolving in self-host cljs running in the browser e.g Klipse and crepl
How can I get the Clojurescript namespace I am in from within a clojurescript program? I want to do this do provide certain debug information (it only needs to work in dev mode)
Namespaces are not first class in ClojureScript as they are in Clojure. There's absolutely no way to get the namespace at runtime. It is possible to get this information at macroexpansion time if you're not afraid of accessing some the ClojureScript compiler internals. There should probably be an API for this - but we're not there yet.
You can get the name of the current namespace with this trick, which takes advantage of :: creating a namespaced symbol for you in the current namespace:
(namespace ::x)
You probably don't want to use that value for anything, because if the code is compiled the internal representation will change. If you want to live dangerously, then in the browser you can then access the js object that holds the namespace like this:
(reduce (fn [ns n] (aget ns n))
js/window
(clojure.string/split (namespace ::x) #"\."))
During macro-expansion you can access &env and retrieve namespace information from the :ns key like this:
(:ns &env)
(:name (:ns &env))
This only works at macro-expansion/compile time and not at runtime.
You can try this
(apply str (drop-last 2 (str `_)))