oxalorg / 4ever-clojure Goto Github PK

• Configure such that an error is thrown for restricted functions, probably by using sci deny
• Show a warning for restricted functions - #14 solves this

I don't know how my younger brother ended up with (]) in the field but as it's there he can't delete anything using backspace on mobile phone. So he has to cut or reload the page.
This is related to #64 because it's the same buggy paren deletion prevention.

This is the reason why I don't use plugins like parinfer and other auto paren balancers. They do their job but then they interfere very much.

I find Character/isUpperCase is not working(some solutions use it though).
I wonder is every java interop not acceptable in the editor?

(fn [s]
    (apply str (filter #(Character/isUpperCase %) s)))

In problem 87:

(= (__ 3 4 5 60) '(= (+ (* 3 4) 5) 60))

should be:

(= (__ 3 4 5 60) '(= (* 3 4 5) 60))

and the last two:

(= (__ 1 2 10 100 2001) '(= (+ 1 (* 2 10 100)) 2001)
(= (__ 1 2 10 100 1300) '(= (* (+ 1 2 10) 100) 1300)

are missing an ending parenthesis

(range) or (iterate f x) will cause a frozen page.

Currently if you open the 'view solutions' of problem 1 - it goes into a frozen state and then after a minute or so it shows up all solutions. Will be good user experience to load first 5-10 solutions and then on scroll down bring the next 10 etc.

https://4clojure.oxal.org/#/problem/1/solutions

Also, can think for creating a Set and remove duplicate solutions. Look at the screenshot with duplicate solution for problem 1 (it might be the case for other as well). Better will be to clean the solutions data one time (remove duplicate) and store it.

When trying to access I'm getting:

This site can’t be reached4clojure.oxal.org took too long to respond.
ERR_CONNECTION_TIMED_OUT

Can you please check it out?

I might take a look at this myself, probably SCI could help in exposing the "pretty" stacktrace from babashka and scittle.

@humorless has an amazing idea to use a chrome extension to scrape individual data from 4clojure.com

https://twitter.com/humorless/status/1413907323139936257

I don't know how reliable it is given the fact that 4clojure.com keeps returning 502

But for folks who want to preserve their solutions it's a great idea.

And then we can add a file select button which takes the csv and imports the data into our localStorage.

I can input * 2 and pass

Why not add this into the description of the problem: https://clojuredocs.org/clojure.core/fn
And then you could add a test that checks for (fn? __)

Currently in home page just above problems list there is

Problems (151)

Perhaps the line could report as well the number of totally passed problems like

Problems (151, passed 38)

Description

Getting a wrong feedback of oh-no although the solution is correct to the current task, after solving more than one task,.
It thinks that the solution is still the the solution you wrote to the previous task.

To Reproduce

1. Answer correct on a question and clicking run to check the solution, you will get complete!.
2. Move to the next question

• The previous answer automatic be shown on the window where you type the answer.

3. Write the correct answer to the new question and run the solution.
4. You will get an oh-no, which means that your answer is wrong although it is correct.

this happens both when navigation by changing the url and using the popup of the solution.

Work around

A hard reset to the browser every-time you move to a new question.

I was messing around with (Variadic Functions) in 4ever-clojure's editor using the example but getting errors

The display of the new editor will break if having too many lines.

Here is my solution to problem 29 - Get the Caps:

(fn [s] (apply str (filter #(<= (int \A) (int %) (int \Z)) s)))

Although it is even listed in the Solution page, it does not pass any test. Is this a bug?

Message Method toUpperCase on function String() { [native code] } not allowed!

In https://tryclojure.org/ I decided to just use*print-length* to prevent (range) from hanging the browser. I think this is nicer and does not rely on hacking SCI. SCI 0.3.1 changed significantly and the hack will not work there anymore (it's still possible but the hack needs to change).

Hello and thanks for keeping this amazing exercise collection online!
Would that be too much of a nuisance to add a counter of one's solved problem somewhere on the main page?

When I try to solve anything on mobile then modal takes the wrong part of the screen.
Why do you want to have that modal at all? Why not display the result inline as the original 4clojure did? It simply worked.

For instance in this screenshot (from my phone) I don't see the result and I can't close the modal because the button it outside of the screen.

This also makes so that you can't simply run commands in the REPL and only when you're happy then you send them to evaluate for the test check. This also hides the results so that you won't be able to see them when you type your code.

Screenshot:

It's not 2002. Llet's show some style, some sugar, and a bit of sass!

https://4clojure.oxal.org/#/problem/16

#(format "Hello, %s!" %)

is one of [the accepted solutions](https://4clojure.oxal.org/#/problem/16/solutions) but I see the following error. 

#error {:message "Could not resolve symbol: format", :data {:type :sci/error, :line 1, :column 7, :file nil, :phase "analysis"}} 

It works on regular bb. 

Babashka v0.3.1 REPL.
Use :repl/quit or :repl/exit to quit the REPL.
Clojure rocks, Bash reaches.

user=> (format "Hello, %s!" "Kenji")
"Hello, Kenji!"

Is it possible to run the tests against the solutions archive?

with problem 100 https://4clojure.oxal.org/#/problem/100 when I run my solution locally I get all tests passing but on the site only 2/5 pass
when I submit any of the solutions from the solutions archive they also only get 2/5 passing so I think something's wrong with the site's evalutation

Alternative is shown at the bottom of the screenshot.

Maybe it's because I'm anglo-centric; but I feel like positive rotation should be clockwise and negative rotation should be anti-clockwise

scraping by Chrome Extension listly.io
LISTLY_new.xlsx

I think the "flow" of solving problems can be improved by showing the link to the next unsolved problem, after solving a problem. Or maybe already even before, so you can just navigate to the previous, next and perhaps a random next problem?

• do this after we start saving user solution state
• make the problem list sortable based on solve status

There are a lot of double solutions here:

https://4clojure.oxal.org/#/problem/2/solutions

Perhaps the front-end could deduplicate.

• line numbers would be cool
• code highlighting? even better
• auto indenting? or maybe indents with tabs

Yeah linting won't work as we do fill in the blanks 🙈

Certain problems incorrectly pass with empty inputs, for example Problem 2. If I run the default empty input on Problem 2, I get a successful completion. This exists for several other problems in the lower-tiers but doesn't seem to be a problem in the higher tiers.

• hosted this on https://github.com/oxalorg/4clojure-solutions-archive
• setup CNAME
• changed URL to load solutions from 3d8db43
• fixed CORS errors in 3b14677

This is on mobile. So there is no Ctrl+-> and delete keys. Also to do character selection I need to use my finger and somehow select a single character.

So not being able to use my backspace gets in the way. Paren matching is great effort but for me it brings frustration only.
For instance let's consider this case:

I'm testing my function to increase a number 5 because I don't yet want to submit the solution because it's not yet correct.
So what I would do is to wrap it into a (__ 5) so that I could test it in the REPL:

((fn [x] (inc x)) 5)

Then I would click Run, close the modal and try to see the result (when I'll find it) (that's also pretty bad).

And then when I want to clean it up (the outer invocation parens) I can't and that's especially bad on mobile.
If I place my cursor here and press backspace then nothing happens and instead it moves to the left by one symbol:

((fn [x] (inc x)) 5)[CURSOR]
->
((fn [x] (inc x)) 5[CURSOR])

So um... that's not good. And then if I want to delete some forms inside I also suffer from this behavior because "unbalanced parens bad".
For me this feels more like I'm forced to do it this way and it's especially bad when I really want to edit very large functions on mobile when I'll do the harder problems.
IMO this hinders the use of this app as only the trivial problems will be attempted at all. Because others will be unusable.

So in that case I end up inserting identity everywhere but that's only useful for small problems. And this is not good (I did do this on original 4clojure Android app but this paren behavior would force me to use it even more).

Why not toggle this? Maybe later on people will understand that they shouldn't delete parens randomly.

Goal: I want to submit my exercise answers w/o using a mouse.

Issue: I am unable to Esc or Tab out of the code text-area to the "Run" button to submit my code answers. It might seem that the focus is stuck in the code text-area... but I have the ability to "Tab-Shift" back out in the opposite direction. Is there a keyboard shortcut to submit my exercise answers?

Proposed Solution: Not sure exactly how atm, but, to enable keyboard only navigation and usage of 4ever-clojure. Accessibility fixes and/or keyboard shortcuts could be possible resolution routes.

Would it be helpful for me to investigate potential fixes?

Thank you guys for making 4Clojure continue on!

Ps. I also noticed that the successful exercise submission modal does not trap user focus as it should - this will make the site less accessible for differently abled visitors. There are a few ways to fix this that keep dev work to a minimum while offering maximal accessibility - I'm happy to support as I can!

1. Write ->> in a solution (or a valid usage of it-- doesn't matter).
2. Press Run.
3. Reload the page or navigate back and forward.

->> becomes -

The original from 4clojure.com is this:

(= (str '__) (__))

The current one expects a function of two arguments and is badly formatted too:

(= (__ '(fn [x] (__ x x))
                            '(fn [x] (__ x x)))
                       ((fn [x] (__ x x))
                        '(fn [x] (__ x x))))

And it means that = is allowed in the new "complicated" version and not in the original version.
Does this mean that all problems are now different?

It was hard in the original, now it's trivial as one symbol will solve it.

Going to next problem doesn't reset state properly

The state is already getting a bit messy. I've started a port to re-frame to make this more manageable. Might as well do it now, before it gets too late.

I'm also adding a better way to handle css.

When evaluating that, the "maximum number of elements reached" is triggered. I believe this is why the exception must be in a lazy seq as well.

(= false (__ [[:a :b] [:a :b] [:a :c] [:c :a]
               [:a :d] [:b :d] [:c :d]]))

So, I am not actually sure what to make of this. I interpreted "unidirectional" to mean that [:a :c] and [:c :a] were the same edge. But if I disregard the two duplicate edges, it seems to me that (d c) (c a) (a d) (d b) (b a) is a valid solution to this. Is the intent that this one is necessarily false because some edges appear twice in the list, and you can't visit them all while only visiting them exactly once, or am I missing something else?

Here are the solutions that I used for 30 and 31 respectively

30:
#(map first
(partition-by identity %1))
31:
#(partition-by identity %1)

I know that the exercises are all very freeform, but I do think that each problem has a few highly idiomatic solutions (and for these 2 I think the ones I've provided are idiomatic); Since problem 30 builds off of problem 31, I think their order should be switched.

Thank you so much for the 4clojure project!

The UI check or eval function is bad because on error (let's say we call ([inc] 1 2 3 4 5)) the custom eval returns a string:

(defn eval-string [source]
  (try (sci/eval-string* context source)
       (catch :default e
         (with-out-str (error-handler source e)))))

This is the code from current version.

Repro:
Go to https://4clojure.oxal.org/#/problem/38 (it will call user's code with multiple args)
Enter [] into the field
Outcome: pass fail pass

A quickfix of if (true? passed?) works when presenting the data but this probably needs to be decided in a better way.

It would be a basic refactoring into some kind of a map or similar thing but this function is used all over the place.

Description:
Incorrect numbering of exercises or missing exercises.
If you go over the numbers of all the exercises, these are the exercises / numbers that are missing:
109, 113, 123, 129, 130, 133, 136, 138, 139, 142, 149, 151, 152, 154, 155, 159, 160, 163-165, 167, 169, 170, 172, 174-176, 178-194

I'm not sure if this is a bug exactly but it looks weird,

This seems to happen when clicking 'Run' and there isn't a popup for tests passing or failing.

The data in data.cljc is effectively JSON / EDN. We don't need to load all this data in the first page load, but only the problem titles. Once you navigate to a problem, you can then load the problem data and show some other solutions when the problem is solved (these solutions could be contributed through PRs). I propose to store every problem in a separate JSON file that can be fetched using an HTTP request to implement this lazy loading.

In problem 119:

(is (= (__ :x [[:o :e :e]
the check of the solution fails completely because the is-macro isn't available.

Since this additional step is not needed, it should be omitted.

Edit:
As a workaround I added a definition for it at the begin of my function as a dummy:
(fn [p b] (defn is [arg] args) (...

Eg. problem 113 and many more

Look into it

https://groups.google.com/g/clojure/c/ZWmDEzvn-Js

For Question https://4clojure.oxal.org/#/problem/16

one of possible solution is

(fn [x] (format "Hello, %s!" x)).

but this does not passes

When you answer a question and then use the link to go to the next question, your previous answer stays in the answer box.
Even after deleting it and filling in the correct one, the answer isn't accepted.
Refreshing the browser fixed it.
I'm on Firefox.

Looks like there was some code introduced that removed the updated code input editor.

I'll submit something to get that back