D.vr is not just a standard object, it has properties that are defined with Object.defineProperty( with getters and a setters. I need a strict equivalent, ie use an assignement to call the proper getters when needed
@GrosSacASac Would Object.assign help?
It uses the getters and setters:
The Object.assign() method only copies enumerable and own properties from a source object to a target object. It uses [[Get]] on the source and [[Put]] on the target, so it will invoke getters and setters.
Of course, that still won't work if you're worried about overwriting existing data
I tried Object.assign but it s not the same. what I need here would be more Object.assignDeep becuase Object.assign goes only 1 level down. Maybe I can use Object.assign instead of the innermost loop
```
let scopeName,
variableName,
DVrScopeName,
commentsDataScopeName;
for (scopeName of Object.keys(commentsData)){
if (!D.vr[scopeName]) {
D.vr[scopeName] = {};
}
DVrScopeName = D.vr[scopeName];
commentsDataScopeName = commentsData[scopeName]
Object.assign(DVrScopeName, commentsDataScopeName);
/* same as
for (variableName of Object.keys(commentsData[scopeName])){
D.vr[scopeName][variableName] = commentsData[scopeName][variableName];
} */
}
can I simplify more ?
This message was deleted
Anyone happen to have a way to get from:
[
[1, 2],
[1, 2, 3],
[1]
]to:
[
[1, 1, 1],
[2, 2],
[3]
]
Kind of rotating the data
I feel like I've seen it before in this room, but I have no idea what the context was anymore.
Did you look at http://ramdajs.com/0.20.1/docs/#transpose ?
@ram-bot
R.transpose([[1,2], [1,2,3], [1]])
[ [ 1, 1, 1 ], [ 2, 2 ], [ 3 ] ]
I knew it! Awesome! Perfect! Thanks! :D
Ramda is so good
As much as es2015 took inspiration in python, coffeescript, I wish new es2017 syntax will be inspired by ramda and co
@GrosSacASac es7
async/await should replace co eventually
I think they might also be introducing Observables
Though I wish they introduced something like a hindley milner type system and typeclasses. That kind of type system can be backwards compatible if it treats a function parameter as an "any" type by default
lool
hindley milner type system and typeclasses yes plz! :fire: :fire: :fire:
@Bradcomp thank you so very much. I don't understand what objectUpdater is doing on your code example. You always seem to have the answer, you have a knack for that style of coding
let objectUpdater = R.compose(R.map(serializeNullValue), updateEntityType)
//Does R.map map over the values of the object returned by updateEntityType?
Yes it does, I answer my own question. @Bradcomp thanks that's smart
Is this something better generalized by another pattern?
//kReduce :: Chain m => (b, a -> m b) -> m b -> [a] -> m b
const kReduce = f => m => xs => xs.reduce((m, x) => m.chain(acc => f(acc, x)), m);
The reason I'm asking is because it seems very useful, but I haven't seen it anywhere else, leading me to conclude that there must be some more widely employed way to solve the same problem.
@Avaq what does it do?
let arrayOfAttributeNames = [null, null, null, null];
let verifyIfAllNil = (acc, next) => {
return ((acc === true) && (next === null)) ? true : false;
}
R.reduce(verifyIfNil, true, arrayOfAttributeNames);
is there a better way do to this?
@ram-bot
R.all(R.isNil) ( [null, null, null, null] )
true
damn
6ewis This message was deleted
@Avaq thanks
not related to ramda but does anyone use proxyquire ?
Is it possible to stub a function from the same module passed to proxyquire ?
const a = () => {}
const b = () => {}
const c = () => R.pipe(a, b)c but without calling the real a and b
Sorry I'm a noob at testing, but I want to improve ! :smile:
@jgoux we banned proxyquite from our codebase
@jgoux we even added a test such that linting fails if additional uses of proxyquire are introduced
wow
xD
why ?
and what do you use instead ?
Quick question, can a curried version of zipWith be used in a transducer function?
Like, I have an array that I'm map'ing over and then I'm calling zipWith immeidiately after
@jgoux the fact you reach for proxyquire implies that you want to change internal behaviour - which is undesirable
instead you should use dependency injection - i.e pass in the dependencies
@gilligan I use proxyquire to mock things like database or ssh
yes, that does not change the fact that you should be passing in a function that carries out an IO operation instead of having the unit that you want to test access the Database and fiddle with proxyquire to overwrite that
when using proxyquire to mock away internals of your unit, your test depends on implementation details
I'm not sure I follow
this is the piece of code I want to test
@gilligan Could you point me what's wrong with it ? ^^
At first I wanted to test only
main as it was the only "public" function
I tried this and it didn't work T_T
let a = [0, 1, 2, 3, 4, 5, 6, 7]
let b = [7, 6, 5, 4, 3, 2, 1, 0]
let transducer = R.compose(R.zipWith((x, y) => { return x * y; }, b), R.map(R.add(2)))
R.transduce(transducer, R.flip(R.append), [], a)Getting....xf.@@transducer/step is not a function
@LeonineKing1199 I don't think
zip and family can act like transducers. Usually the documentation would state that explicitly.
Nuuuuuuuuu!!!!
@jgoux sorry, busy now.. it boils down to not passing in
dbConnection but a function that will return the result of a database operation - the real code will carry out actual db interaction - the test code will just pass in a function that returns the result
this is why for testing "main", I'd need to stub getPatient + assertPatient + getLastDemandesEditables no ?
so I don't care about their implementation ?
Then I can test one level deeper with for example getPatient
@jgoux What gilligan is trying to say is that you means of stubbing is flawed. You're trying to stub by hacking your implementation, whereas you could stub by "injecting" the things you'd want to stub. So for example
main would take getPatient, assertPatient and getLastDemandesEditables as arguments. Then from your test you could just call main with stubbed versions of those dependencies.
oh ok :D
But the function signature would be...weird
so it would be better to always inject modules + internal function explicitly ?
I can always export a partially applied function :p
I think it's generally good practice bc it leads to less highly coupled code, so you can decompose it into unit tests, which would make you less reliant on e2e tests and less likely to have merge conflicts when different devs work on different parts of the frontend, and easier to debug
yes I'm trying to adapt my code right now to inject all the dependencies ^^
this is really nice : https://github.com/DrBoolean/freeky#custom-types
But I'm a little afraid of monads in JS without a type system
Don't do monads in JS, do monads in C++!!!!
http://bartoszmilewski.com/2011/07/11/monads-in-c/
This guy is honestly one of my favorite bloggers XD
Yes, I love this serie : http://bartoszmilewski.com/2014/10/28/category-theory-for-programmers-the-preface/
I love that guy :)
Bringing functional programming concepts to a language people think only supports OOP. Seriously, thank God for std::function and closure types.
But I'm a little afraid of monads in JS without a type system
My too, @jgoux, which is why I use a (run-time) type system. ;)
Which one ?
sanctuary-def ?
I think I'll try to use one too
Yeah. Either directly or indirectly via Sanctuary.
It has caught bugs in production code with 100% test coverage!
what do you mean by directly or indirectly ?
I don't necessary use sanctuary-def in every project, but because Sanctuary uses it I get the benefits whenever I use Sanctuary functions and types.
A mistake I often make is mixing monads.
I'll have a value of type
Either String String and will want to convert the Right to an integer.
I might do something like
R.chain(S.parseInt(10), e), forgetting that I'm in Either context rather than Maybe context.
Either#chain will complain, though.
yes this is why I don't want to use a lot of them right now, because of the stack it creates, without type checking it's a nightmare
I don't imagine the look of a code using something like : Reader + Future + Either + Maybe
Try evaluating this expression on one of the REPLs on the Sanctuary website:
R.chain(S.parseInt(10), S.Right('42'))
nice ^^
@ram-bot
R.chain(S.parseInt(10), S.Right('42'))TypeError: ‘Either#chain’ is expected to return a value of type (Either a c); returned Just(42)
It makes it obvious what I got wrong and how to correct it.
Yes indeed
I didn't know I can use R.chain to chain monads
xD
I always do something like :
S.pipe([a(), F.chain(b), F.chain(c)])
cleaner with R.chain !
(I used the
chain from fluture)
Nice.
@davidchambers Do you use sanctuary-def on your own functions ?
I certainly do. Check out
generate, easily the nicest build script I've written to date.
Will check thanks !
what is the second argument of
def ?
This gives me warm fuzzies:
pipe([at(2),
maybeToEither('Missing command-line argument'),
map(lift2(pair, I, I)),
map(over(lensIndex(0), version)),
map(over(lensIndex(1), readme)),
chain(ifElse(all(isRight),
pipe([rights, Right]),
pipe([lefts, join('\n'), Left]))),
map(over(lensIndex(0), wrap(' <small>v', '</small>'))),
map(apply(replace(/(?=<[/]h1>)/))),
map(toDocument),
chain(writeFile('index.html')),
either(failure, success)],
process.argv);
The second argument to
def indicates the type-class constraints.
@davidchambers I don't understand the $.env initialization. You register all the types you will check ?
what if I make a type
Point and I don't put it in the $.env ?
or is this just a shortcut for accessing $.Point ?
const X = curry(function(f, g, x, y) { return f(g(x,y)); });
const isFirstOrNone = X(gte(0), lastIndexOf);
const GFirstOrNone = isFirstOrNone('G');
GFirstOrNone('G456546');//-> true
GFirstOrNone('5G456546');//-> false
Is there any common Combinator or pattern I'm missing a cleaner abstraction for implementing something like the above (basically, the rightmost function needs to get two arguments, preferrably one by one, before it's result can be passed to the leftmost one, so just
compose(gte(0), lastIndexOf) wouldn't quite work)?
node.js 4.3 support finally added to aws lambda:
https://forums.aws.amazon.com/ann.jspa?annID=3672
https://forums.aws.amazon.com/ann.jspa?annID=3672
does someone have an idea how I can make this... more functional? also my structure doesn't match 100% what I want (included).
I was wondering if I could use R.lensIndex with R.lensPath but R.lensPath(path) creates objects instead of an arry.
with
R.set
(e.g. http://goo.gl/TEqZ2H)
@dtipson
const isFirstOrNone = R.pipe(R.head, R.equals('G'));
isFirstOrNone('G456546');//-> true
isFirstOrNone('5G456546');//-> false
not exactly an answer to your question, but this does the same job I guess.
maybe take a look at
R.partialRight
I don't understand the $.env initialization. You register all the types you will check ?
I agree that this is somewhat confusing, @jgoux. The reason it's necessary to define an environment (a list of types) is to support type variables. In order to determine whether [x, y, z] is a valid value of type $.Array(a), we need to know all the possible values of a – in other words, all the types we'd like to consider.
For a type such as $.Array($.Number) we don't need knowledge of the environment.