See: https://github.com/ericelliott/rfx

Since we're in favor of codeclimate, to see it's full potential and to increase sanity and simplicity for our current and future projects, we should pick a JavaScript style. CodeClimate offers a list of eslint plugins that are supported, and the ones that seem to be relevant for this purpose are:

• airbnb.
• google.
• standard.

There are also many other benefits of using standard stuff (pardon my laziness 🍭).

I can't fix it since it's asking for an admin to do it 😶

See https://runkit.com/daedalus28/splitby-futil-candidate

Every real library has a fancy logo. We clearly need one.

Maybe something looks like a borg cube?

• Add a link to the changelog in the readme
• Set up CI to auto publish to npm
• Set up CI to auto run unit tests
• Document all available methods

Hi! Just as a reminder :) I don't want to forget that.

For the repeated array thing

Version 2.4.0 of eslint-plugin-import just got published.

This version is covered by your current version range and after updating it in your project the build failed.

As eslint-plugin-import is “only” a devDependency of this project it might not break production or downstream projects, but “only” your build or test tools – preventing new deploys or publishes.

I recommend you give this issue a high priority. I’m sure you can resolve this 💪

Your Greenkeeper Bot 🌴

The goal is to capture this use case for radio buttons and mutex checkboxes:

{...{
  onChange: e => { field.value = e.target.value },
  value: option,
  checked: field.value == option,
}}

Given that we have checkboxValues, we probably shouldn't call it checkboxValue as that'd be too close. Likely name would be radioValue which is probably more intuitive than mutexValue

I think it can be like this:

export let asyncFlow = (head, ...tail) => async (...params) => asyncFlow(...tail)(await head(...params))

The should support .onError and before (concurrency, errors, and status aspects) and should support single error unit tests are failing.

They are the same :)

processQuotes may not be needed, consider removal

Add an uncurry fn

Example
(n, f) => _.curry(R.uncurryN(n, f))

I am trying to use this library in a Typescript project (expecially the function helpers are great in my case). However, the project itself includes no .d.ts file and there is also none via npm install @types/futil-js.

Would it be possible for the author to create/add a typescript definitions file?

https://smartprocure.github.io/futil-js/#api-lang-append

Given the order of concatenation is reversed from the order of the arguments, this would be better described as prepend instead of append.

F.postings(/a/g, 'vuhfaof') returns [ [ 4, 5 ] ]
F.postings(/a/, 'vuhfaof') never returns at all -- you're stuck in the while loop at https://github.com/smartprocure/futil-js/blob/master/src/regex.js#L42-L45 because without the g flag regex.exec() doesn't consume the initial match and so just finds it again forever.

Two possible fixes:

• add the g flag before the while loop when needed with regex = new RegExp(regex.source, regex.flags + 'g')
  • downside: doesn't really give the behavior that would be expected by a person who on purpose didn't use the g flag. They're expecting the first match.
• check for the g flag. if it's not present, just return the rest of .exec-ing, else do the while loop
  • or equivalently have the while loop look at a boolean that you flip when you enter it.

Either fix is simple but I'll wait on feedback as to the preferred behavior.

Browser tests are still failing. See: https://circleci.com/gh/smartprocure/futil-js/968?utm_campaign=vcs-integration-link&utm_medium=referral&utm_source=github-build-link

Here's a plain text copy of the issue:

Chrome 26.0.1410 (Windows 7 0.0.0) ERROR
  {
    "message": "Uncaught ReferenceError: __VERSION__ is not defined\nat webpack:///node_modules/lodash/lodash.min.js:5:1 <- src/index.js:5",
    "str": "Uncaught ReferenceError: __VERSION__ is not defined\nat webpack:///node_modules/lodash/lodash.min.js:5:1 <- src/index.js:5"
  }
Chrome 26.0.1410 (Windows 7 0.0.0) ERROR
  {
    "message": "Uncaught ReferenceError: __VERSION__ is not defined\nat webpack:///node_modules/lodash/lodash.min.js:5:1 <- src/index.js:5",
    "str": "Uncaught ReferenceError: __VERSION__ is not defined\nat webpack:///node_modules/lodash/lodash.min.js:5:1 <- src/index.js:5"
  }
Chrome 26.0.1410 (Windows 7 0.0.0): Executed 0 of 0 ERROR (1.06 secs / 0 secs)

We should have a real docs site like lodash, ramda, etc.

Currently, we're waiting on lodash 5 progress since we'd like to do the same thing as lodash, but their doc generation is currently unclear while they're working on v5.

This seems to be reproduced in many browsers:

Mobile Safari 8.0.0 (iOS 8.4.0) ERROR: 'Unhandled promise rejection', AssertionError{message: 'expected promise to be rejected with 'Error: hi from aspect' but it was rejected with 'Error: hi from aspect'', showDiff: true, actual: 'Error: hi from aspect', expected: 'Error: hi from aspect', stack: 'AssertionError@http://localhost:9876/base/test/aspect.spec.js?45512c279a5a25377ccf6505f37f502ef315c0bc:2681:22

In #35 we picked standard js style. The the PR in #30 made us conform but we need to add our hard earned shiny readme badge 📛

There are underscore/lodash decorators that might come handy, such as:

• allong.es
• lodash-decorators

The defaultValue for cascade seems like a bit of a hack. Instead, there should probably be two versions, one without the default value named cascade and one with the default value named cascadeOr. The latter would follow the argument order of getOr, that is, default value first.

arrow functions does not binding its own arguments,

• need to change the signature of maybeCall(fn) to maybeCall(fn, ...args) in order to call fn(...args) inside.

• add test for passing arguments to fn() call inside maybeCall

Using _ as a placeholder obviates the need for the In method conversions, because they only save a single character at the cost of new things to learn:

F.someMethodIn(x) vs
_.someMethod(_, x)

Additionally, sometimes our conversion methods would be useful together, e.g. an In and an On. A good example is unset - you might want to unset both in and on an object, and defining compound conversion conventions seems messy (unsetIon?).

In order to make this work, we either need to explicitly curry all our conversions, or wait for lodash/lodash#3440 to be fixed

This is a followup to #144 to make sure that kind of thing never happens again.

As part of CI, we should run all of our unit tests in every browser. Sauce Labs and Browserstack free for open source so there's very little friction for us to do this.

This a method to walk a tree with async visitor/iteratee methods

We can get them all here:
https://libraries.io/npm/futil-js/versions

It seems like our master build is failing CI. Locally, it seems like it's due to something with eslint-plugin-mocha from eslint-config-smartprocure 🤔

@sadasant any ideas?

Should we move lodash to peer dependencies for futil? I'm wondering if that's adding extra weight to our downloads.

I know that we've been running tests against the lib file to make sure that the build works, but since this is the responsibility of babel, we could focus the tests on our source code instead. Among the benefits, we could perceive: an overall simplification of the npm scripts, cleaner test coverage and faster tests.

We use this sort in several places, let's add it here:

let sortByKey = _.curry((key, tabs) => _.sortBy(obj => {
  let ord = _.get(key, obj)
  return ord >= 0 ? ord : ord < 0 ? (tabs.length * 2) + ord
    : tabs.length
}, tabs)

Helps with determining what version is really running, which occasionally is a mismatch with package.json (e.g. if versions are not locked down)

If we want to use CircleCI, we need to start it from a github organization owner.

Thanks to the awesome @joaonuno, we now have the futil package name!

This ticket is to track all the steps we need to take to make it happen

• Rename this repo to smartprocure/futil
• Publish a new version of futil-js with the new repo name and a big deprecation warning at the top of the readme (and ideally with our fancy new logo to keep brand identity consistent)
• Change package name and release a new version (major or feature? @sean-c is our resident semantic-versioning expert)
• Run npm deprecate on the futil-js version
• Update all dependent libraries/projects
• Enjoy the awesomeness of our new package name

Version 3.5.0 of webpack just got published.

This version is covered by your current version range and after updating it in your project the build failed.

As webpack is “only” a devDependency of this project it might not break production or downstream projects, but “only” your build or test tools – preventing new deploys or publishes.

I recommend you give this issue a high priority. I’m sure you can resolve this 💪

Your Greenkeeper Bot 🌴

A trampoline is a loop that iteratively invokes thunk-returning functions (continuation-passing style). A single trampoline is sufficient to express all control transfers of a program; a program so expressed is trampolined, or in trampolined style; converting a program to trampolined style is trampolining. Trampolined functions can be used to implement tail-recursive function calls in stack-oriented programming languages.–Wikipedia

See this post: http://raganwald.com/2013/03/28/trampolines-in-javascript.html

My dear friend @stefanmaric wrote a trampoline in this Gist: https://gist.github.com/stefanmaric/1dfc01dd3985d5b423f22fcad7c58da2

const trampoline = (fn) => (...args) => {
  let result = fn(...args)
  while (result instanceof Function) result = result()
  return result
}

The idea is to enable shorthand so a scenario like this:

F.findApply(x => x.b, [{ c: 1 }, { b: 3 }]) // 3

Could be shortened to this:

F.findApply('b', [{ c: 1 }, { b: 3 }]) // 3

Just requires wrapping f in _.iteratee in collection.js

It'd be useful for quickly browsing the documentation if:

1. Functions were sorted alphabetically
2. There was a search box to quickly zero-in on a given function.
3. Documentation could be exported to different formats for offline viewing (PDF, Markup, etc...)

Similar to #138, if we instead construct 1 regex, it will be much faster. The match any case is easy - just join words with |.

We should improve our runkit example to require futil as F, and maybe add one or two small examples.

http://blog.runkit.com/2015/10/28/npm-plus-runkit/

See:

• services/PromiseUtils.js
• servicesutil/SearchProcessor/utils.js

https://runkit.com/daedalus28/basic-mongo-json-schema-validation

The checks are domain-specific. The validateTree function could be a good futil helper, although its API needs to be discussed. Do we return a flattened tree? Do we give the option to bail on first error? Etc...

And then importing from index so that we can do the dynamic reduce thing but still have them as top level exports

Not just useful for a docs site, exploreAPI (currently https://runkit.com/daedalus28/explore-api-futil-candidate) could be used as part of a danger check to see if newly added API method's test cases match existing methods and warn if they do.