remove coffeescript dependency and use vanilla .js files

[kaizhu256]

this pull-request is broken up into 4 commits:

1) create vanilla .js files from .coffee by running shell-command:

# revert individual .coffee files to .js
npm install [email protected]
for LIB in api-data api exports worker
do
    cat "src/$LIB.coffee" | npx coffee --bare --compile --stdio > "src/$LIB.js"
done
# restore package-lock.json
git checkout package-lock.json

2) update Makefile to use vanilla src/.js files instead of src/.coffee

3) remove references and dependencies to coffeescript in files:

$ git grep -i coffee
.github/workflows/CI.yml:    - name: install coffee
.github/workflows/CI.yml:      run: npm install --global coffeescript@1
.github/workflows/release.yml:      - name: install coffee
.github/workflows/release.yml:        run: npm install --global coffeescript@1
.npmignore:coffee/
package-lock.json:              "coffeescript": {
package-lock.json:                      "resolved": "https://registry.npmjs.org/coffeescript/-/coffeescript-1.12.7.tgz",
package.json:           "coffeescript": "^1.12.7",

4) remove .coffee files:

deleted:    src/api-data.coffee
deleted:    src/api.coffee
deleted:    src/exports.coffee
deleted:    src/worker.coffee

[lovasoa]

Hello and thank you for this PR !
I am not against moving away from coffeescript. The decision to use it was taken several years ago, when javascript was not what it is today.
But we can't keep the raw converted files, which are not human-readable, maintainable javascript. We'll have to clean up the :

var _, binaryDb, func, ref, ref1, stmt;
    ref = this.statements;
    for (_ in ref) {
      stmt = ref[_];
      stmt['free']();
    }

or

        data_func = (function() {
          switch (false) {
            case value_type !== 1:
              ...
            case value_type !== 4:
              return function(ptr) {
                var blob_arg, blob_ptr, j, l, ref1, size;
                ...
                for (j = l = 0, ref1 = size; 0 <= ref1 ? l < ref1 : l > ref1; j = 0 <= ref1 ? ++l : --l) ...
                return blob_arg;
              };
            ...
          }
        })();

And since manual work will be required on all of the source files, I thought we may as well add type information and switch to Typescript. Would you be interested in such a project @kaizhu256 ?

[lovasoa]

The port to typescript was already implemented in #290 by @pwnsdx

[kaizhu256]

And since manual work will be required on all of the source files, I thought we may as well add type information and switch to Typescript. Would you be interested in such a project @kaizhu256 ?

i'm not really a typescript fan ^^;;; the intent of pull-request is to make project more accessible to people put off by excessive tooling (including typescript).

i'll be happy to refactor the .coffee files to [mostly] pass jslint. yea, i kno jslint is tooling as well, but at least its human-readable vanilla-js.

[lovasoa]

OK, let's first clean up the javascript, we'll then see about whether we want to add type information.

[kaizhu256]

add 2 more stable commits (6 total):

commit (5)

simplify ci-build and vanilla-js cleanup by merging following files into api.js, respectively:

        src/output-pre.js
        src/api.js
        src/exports.js
        src/api-data.js
        src/output-post.js

commit (6)

rename OUTPUT_API_FILES to SOURCE_API_FILES in Makefile

[lovasoa]

If our code passes jslint, I think we should add jslint to the devdependencies and to the tests

[lovasoa]

Ok, I think we are starting to see clearer. There are still a few coffeescriptisms to remove and we'll be done.
Especially, a lot of loops can be simplified, ref variables can be inlined, and var keywords can be put directly where the variables are declared.

[kaizhu256]

add stable commit # 7 (eccffee)

add file test/jslint.js and npm-command "npm run test-jslint"
todo - allow test-jslint to fail after we finish refactoring src/api.js (test-jslint currently always pass)

add stable commit # 8 (809a47f)

lint src/api.js - resolve conversation - we need only 1 variable
lint src/api.js - resolve conversation - Object.values()
lint src/api.js - resolve conversation - we can iterate in increasing order
lint src/api.js - resolve conversation - the var could be directly in front of the variables
lint src/api.js - resolve conversation - remove merge-comments
lint src/api.js - unindent case-statement
lint src/api.js - use jslint-style-guide of quoting strings with double-quote instead of single-quote
lint src/api.js - hoist constant vars to top of file
lint src/api.js - rename property SQlite.xxx to var SQLITE_xxx
lint src/api.js - remove var runCompiledCode
lint src/api.js - use 4-space indent

[kaizhu256]

add stable commit # 9 (9afe9f1)

jslint src/api.js - fixed all warnings
jslint - "npm test" will now fail if src/api.js fails jslint-check
jslint src/api.js - fix whitespace warnings

@lovasoa, i think this pull-request is ready for merge on my side. can you review when you have time?
also, can we skip refactoring and regression-testing for src/worker.js in a separate pull-request ?

[lovasoa]

Thank you for your fantastic work, @kaizhu256 !
I am going to try and find the time to work on this during this weekend.
I don't think this is ready for merging right now. In particular, we cannot afford to check in and maintain a fork of jslint.
I have access to your branch so I can work on this PR directly, and keep this in the form of a PR until this is ready for merging.

[kaizhu256]

ok and appreciate the thx : ) i'm excited and sincerely believe sql.js is the future of state-management in browsers!