reasonml / reason-tools Goto Github PK

ocamldoc refmt conversion doesn't have syntax highlighting.

When I go to the URL: https://www.tjson.org/spec/, the extension automatically swaps the stylesheet and detects the page to be an OCaml doc page.
This is a false positive.

Every module in this index should be a clickable link. After conversion they're not, and there doesn't seem to be any way to get them back either:
https://ocaml.janestreet.com/ocaml-core/latest/doc/core/

Its converting

let newItem () => {
  lastId := !lastId + 1;
  {id: !lastId, title: "Click a button", completed: true};
};

to

let newItem = () => {
  :=(lastId, lastId^ + 1);
  {id: lastId^, title: "Click a button", completed: true}
};

when maybe it should be:

let newItem = () => {
  lastId := lastId^ + 1;
  {id: lastId^, title: "Click a button", completed: true}
};

See: reason-tools share link

If this doesn't add too much overhead to the build process. There are a few good bindings in here. Low pri though

I get this error when I try to do npm install

npm ERR! code EBADPLATFORM
npm ERR! notsup Unsupported platform for [email protected]: wanted {"os":"darwin","arch":"any"} (current: {"os":"linux","arch":"x64"})
npm ERR! notsup Valid OS:    darwin
npm ERR! notsup Valid Arch:  any
npm ERR! notsup Actual OS:   linux
npm ERR! notsup Actual Arch: x64

This is due to being shrinkwrapped on a mac. Removing the "fsevents" tree from "npm-shrinkwrap.json" seems to work. As would removing "npm-shrinkwrap.json" I suppose, but that would kind of defeat the purpose of shrinkwrapping. I don't know if npm is able to fix the missing fsevents on your machine though.

http://ounit.forge.ocamlcore.org/api-ounit/index.html uses with nested
instead of

. We should respect this choice and be able to add inline 
s into the result.

Unless there's a new selection ofc.

For the following cases:
- Webpages that aren't converting properly

• Webpages that should be converted but aren't
• (Eventually perhaps) webpages that are successfully converted manually.

(Then we just need to devise some machine learning algorithm to intelligently detect and convert the pages based on "big data", and we're home free!)

Similar to OCaml documentation - lets do https://realworldocaml.org/ automatically so that learning Reason is painless.

Which kinda sounds like a good thing, but it's really not. See this example:
http://ounit.forge.ocamlcore.org/api-ounit/index.html

The code listings here use <code> elements, which are really inline elements (but who cares about semantics, it's not like these'll ever be interpreted by a machine, right?). They are however surrounded (not wrapped) by empty <pre> elements (WTF?). This can probably be used to detect these special snowflakes and replace the <code> elements with <pre>s.

http://erratique.ch/software/react/doc/React

There are way too many of them.

Not just for toggling off an autoconverted page, but to try conversion of one that isn't detected properly. E.g. like this page, which I think should be converted successfully if attempted:
http://caml.inria.fr/pub/docs/manual-ocaml/extn.html

It would be nice to have error messages other than just "Syntax error on line 1".

Right now Reason uses JS grammar highlighting. It works OK for impl - but really sucks for types. I tried using https://github.com/patrick-steele-idem/codemirror-atom-modes, with the official reason grammar, but things seemed to get worse - I think there needs to be a manual codemirror mapping.

It'll be better if formatting for pipe chaining will be like this

let prt_elems elems =>
  elems
  |> String.split on::':'
  |> List.dedup compare::String.compare
  |> List.iter f::print_endline;

than this

let prt_elems elems =>
  elems |> String.split on::':' |> List.dedup compare::String.compare |>  List.iter f::print_endline;

After I run npm build I got this:

>tree _build
_build
├── ocamlfind
│   ├── bin
│   ├── doc
│   ├── etc
│   ├── lib
│   │   ├── findlib.conf
│   │   └── ocaml -> /Users/bobfang/projects/reason-tools/node_modules/ocaml/lib/ocaml
│   ├── man
│   ├── sbin
│   └── share
└── refmt
    ├── app.byte
    ├── app.js
    ├── app.map
    ├── reasonToolsRefmt.cmi
    ├── reasonToolsRefmt.cmo
    └── reasonToolsRefmt.cmt

And there is no error message. Any reason you can guess? Thanks!

Low pri, since this only affects contributors. Rebel's install time is pretty brutal.

Port to Safari

An example is the page https://caml.inria.fr/pub/docs/u3-ocaml/ocaml051.html which looks like this for me:

Currently, when I click on the Re button provided by the extension, it displays the same, whether the current syntax is Ocaml or Reason.

I would suggest making the button have some visual indicator of whether the displayed code is Ocaml or Reason.
For example, it might stay highlighted if the code is Reason, and have the current state if the code is Ocaml.

See: http://ounit.forge.ocamlcore.org/api-ounit/

It somehow ends up checking .classList.contains on a text node, which obviously fails.

I'd like to get retop working inside of the popup. Also would be cool to add a dev tools section that looked like the chrome console that could run Reason.

Reason-tools styles this URL as a Reason document: http://nim-lang.org/index.html.

Port to Edge

(Note to self)

Either automatically or with a very easy command would be ideal.

Chrome store wants a single 440x280 and 1-5 1280x800 or 640x400 screen shots. A better descirption would probably be nice as well!

Port to FF

Noticed this after #47b3991 - when I click the stylesheet swap button, I get the OCaml doc stylesheet but the button disappears so I cannot swap back to the Reason stylesheet.

They currently don't autoconvert, and don't render quite right, although it's close. They're based on ocamldoc.

https://bloomberg.github.io/bucklescript/api/index.html

I'd like for this extension to be entirely in Reason. Unfortunately, I'm not super excited about jsoo and I was having trouble setting up bs to work with this project.

I'm trying to build the extension, but I'm getting an error cannot find reason-react:

> npm run build:ext

> [email protected] build:ext /Users/ville/Projects/reason-tools
> eval $(dependencyEnv) && nopam && bsb -make-world


/Users/ville/Projects/reason-tools/node_modules/dependency-env/.bin/dependencyEnv.js:167
    throw err;
    ^
cannot find reason-react in /Users/ville/Projects/reason-tools,/Users/ville/Projects/reason-tools
sh: Failed: command not found

npm ERR! Darwin 16.5.0
npm ERR! argv "/Users/ville/.nvm/versions/node/v7.8.0/bin/node" "/Users/ville/.nvm/versions/node/v7.8.0/bin/npm" "run" "build:ext"
npm ERR! node v7.8.0
npm ERR! npm  v4.2.0
npm ERR! file sh
npm ERR! code ELIFECYCLE
npm ERR! errno ENOENT
npm ERR! syscall spawn
npm ERR! [email protected] build:ext: `eval $(dependencyEnv) && nopam && bsb -make-world`
npm ERR! spawn ENOENT
npm ERR!
npm ERR! Failed at the [email protected] build:ext script 'eval $(dependencyEnv) && nopam && bsb -make-world'.

The problem seems to be that the reason-react version defined in the shrinkwrap file installs a package named rehydrate (the old name of reason-react), so dependencyEnv doesn't find it. I tried to upgrade to the latest [email protected], which uses the new name, but ran into other dependency issues (some inconsistencies between package.json and npm-shrinkwrap.json). A little help getting the project to build would be appreciated.

• Phase 1: Maintain a blacklist/whitelist of url patterns for pages it doesn't properly "detect".

• Phase 2: Add UI to let users add their own url patterns to these lists (with intelligent suggestions)

• Phase 3: Allow users to report/submit the patterns they add to either list, either as they add them, in bulk or as a configurable option to automatically submit (re #32)

The link in the readme points to 404

reasonml-old/reason-react-example@6f2e7b1

Doesn't convert to ocaml and takes away some of the semantic meaning of the page:

http://bloomberg.github.io/bucklescript/api/Js_re.html

The purpose of this issue is to provide a better overview of conversion issues, and to provide test cases, both current and historic to prevent regressions.

Should be converted

• https://bucklescript.github.io/bucklescript/api/Js_re.html (#47)
• https://realworldocaml.org/ (#15)
• https://bucklescript.github.io/bucklescript/api/Js.Global.html (#77)

Should not be converted

• http://asciidoctor.org/docs/user-manual/ (#40, blacklisted)
• http://nim-lang.org/index.html (#21)
• https://caml.inria.fr/pub/docs/u3-ocaml/ocaml-objects.html (#17, blacklisted)
• https://www.mercurial-scm.org/wiki/MergeDriverPlan

Incorrect formatting/styling

• https://mwhittaker.github.io/distributed-systems-ocaml/code_MorePipes.html (#49, no-pre, blacklisted)
• https://ocaml.janestreet.com/ocaml-core/latest/doc/core/ (#46, blacklisted)
• https://ocaml.janestreet.com/ocaml-core/latest/doc/core/Std/Option.mod/ (#27, odoc)
• http://ounit.forge.ocamlcore.org/api-ounit/index.html (#26, no-pre)
• ocamldoc link preservation (#11)
• https://bucklescript.github.io/bucklescript/api/Js.Global.html (#77)

Fails refmt

• http://ounit.forge.ocamlcore.org/api-ounit/ (#42)
• https://caml.inria.fr/pub/docs/u3-ocaml/ocaml051.html (#23)

Current refmt of ocamldoc removes markup, so the internal linking gets lost. ideally we'd keep the correct tokens linked.

With the new hosted refmt page, both the page and popup should offer a share link to it

No idea what RE and </> do

Or some other way of easily converting code from the clipboard.

Ideally there would be a single process watching and we'd have an incremental build flow from Reason into JS.

Right now, Rebel can't watch on OSX and there isn't a great way to integrate webpack and Rebel. Looking into ideas there.

See: https://ocaml.janestreet.com/ocaml-core/latest/doc/core/Std/Option.mod/

Edited to add:

For comparison:

It is much easier to scan the original page for the function you are interested in.

Makes it easy to get a shareable link, and probably even makes local storage obsolete.

This page is shown in all one line, however when toggling the extension, it still has problems with the original text. It looks fine in a browser without the extension installed.

http://mwhittaker.github.io/distributed-systems-ocaml/code_MorePipes.html

False positive. This is the documentation for a static webpage generator, produced by said generator, so this might be producing lots of false positives around the web. But it's also the generator used by Bucklescript, which does not seem to be afftected.

It's also slow as hell to convert because it's one gigantic page, and the swap style button doesn't work on it.

http://asciidoctor.org/docs/user-manual/