-
Notifications
You must be signed in to change notification settings - Fork 0
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Add strict type checking via TypeScript #85
Merged
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
mbland
force-pushed
the
strict-typing
branch
2 times, most recently
from
January 13, 2024 17:37
60c61d9
to
dfddd81
Compare
Adds Typescript to devDependencies and a jsconfig.json files to ensure Visual Studio Code performs more thorough type checking. Description largely copied from: - mbland/test-page-opener#22 mbland/test-page-opener@a63f274 - mbland/test-page-opener#23 mbland/test-page-opener@01a79f6 - mbland/jsdoc-cli-wrapper#20 mbland/jsdoc-cli-wrapper@fafcd21 - mbland/rollup-plugin-handlebars-precompiler#7 mbland/rollup-plugin-handlebars-precompiler@eb5b9a8 - mbland/rollup-plugin-handlebars-precompiler#8 mbland/rollup-plugin-handlebars-precompiler@8b36b2a The code is still JavaScript, but now we get strict type checking in Visual Studio Code and in continuous integration via `tsc` in `pnpm typecheck`. The docs generated by 'jsdoc' are a little funky, and we don't get as much documentation in Visual Studio Code as I expected. I believe I can fix these issues at some point with this foundation in place. The actual changes include: - Added @types/chai, jsdoc, and typescript as devDependencies. - Set .eslintrc to disable the no-undefined-types rule by extending "plugin:jsdoc/recommended-typescript-flavor-error". This is because the Handlebars types in lib/parser.js weren't trivial to replicate, and TypeScript finds those types just fine. This was based on advice from: > ...the config plugin:jsdoc/recommended-typescript-error should > disable the jsdoc/no-undefined-types rule because TypeScript itself > is responsible for reporting errors about invalid JSDoc types. > > - gajus/eslint-plugin-jsdoc#888 (comment) And: > If you are not using TypeScript syntax (your source files are still > .js files) but you are using the TypeScript flavor within JSDoc > (i.e., the default "typescript" mode in eslint-plugin-jsdoc) and you > are perhaps using allowJs and checkJs options of TypeScript's > tsconfig.json), you may use: > > ```json > { > "extends": ["plugin:jsdoc/recommended-typescript-flavor"] > } > ``` > > ...or to report with failing errors instead of mere warnings: > > ```json More background: - https://github.com/gajus/eslint-plugin-jsdoc/blob/main/docs/rules/no-undefined-types.md - gajus/eslint-plugin-jsdoc#99 - gajus/eslint-plugin-jsdoc#1098 - jsdoc/jsdoc#1537 - Added `settings.jsdoc.preferredTypes.Object = "object"` to .eslintrc to enable "Object.<..., ...>" syntax in a JSDoc `@typedef`. Got rid of some extra whitespaces in .eslintrc, too. - https://github.com/gajus/eslint-plugin-jsdoc/blob/b60cbb027b03b4f6d509933b0dca8681dbe47206/docs/rules/check-types.md#why-not-capital-case-everything - https://github.com/gajus/eslint-plugin-jsdoc/blob/b60cbb027b03b4f6d509933b0dca8681dbe47206/docs/settings.md#settings-to-configure-check-types-and-no-undefined-types - Added '.js' extension to all internal imports and added JSDoc comments everywhere reqired by `pnpm typecheck`. - Added 'jsdoc-plugin-typescript' to the build to handle the TypeScript `import().Type` directives. This ended up pulling in the 'es-abstract' module, which blew up the pnpm-lock.yaml file. If I get an itch, I'll implement my own plugin one day and replace it. - Updated `pnpm test-ci` to incorporate `pnpm jsdoc` and `pnpm typecheck`. Added 'jsdoc' to devDependencies to enable this. - Added `null` checks everywhere reqired by `pnpm typecheck`. Added tests to cover all the `null` cases. - Added globals.d.ts and a `/* global STRCALC_BACKEND */` ESLint comment to calculators.js to properly type check `globalThis.STRCALC_BACKEND`. Ironically, this required just referencing it as `STRCALC_BACKEND` without `globalThis`. - Added a temporary components/template.d.ts containing the Handlebars Template() type declaration. Once I properly export those types from rollup-plugin-handlebars-precompiler, I'll remove it. (That plugin currently contains lib/template.d.ts, not types/template.d.ts.) - Added a test for the `#missing app element` case in main.js by adding a new test/main-missing-app-div.test.js. I need to port it to mbland/test-page-opener to solve the coverage problem encountered in: - mbland/test-page-opener#23 mbland/test-page-opener@01a79f6 > - Added a new missing.html and "JsdomPageOpener > doesn't throw if > missing app div" test case to cover new null check in > test-modules/main.js. > > This did, however, throw off Istanbul coverage, but not V8 > coverage. Running just the "doesn't throw" test case shows 0% > coverage of main.js, even though the test clearly passes. My > suspicion is that Istanbul can't associate the > `./main.js?version=missing` import path from missing.html with the > test-modules/main.js file. > > So now `pnpm test:ci:jsdom` will use the V8 provider, and `pnpm > test:ci:browser`, which doesn't use missing.html, will continue to > use Istanbul. Each task outputs its own separate .lcov file which > then gets merged into Coveralls. - Updated `setupFetchStub()` to detect the type of the `body` argument and call `JSON.stringify()` itself if it's an `object`. This eliminated the need for most callers to call `JSON.stringify()`. - Updated `StringCalculatorPage` with typing information and made it so that an empty object will stand in for `null` elements. This is playing loose with typing a bit, as any `null`s will cause errors showing unknown property access. But that seemed better than burdening all callers to do their own `null` checks or workarounds. Of special note: - Added the `instantiate()` parameter to Calculator.init() to inject a Handlebars Template() function. This enabled testing that a missing `#numbers` element was logged by Calculator.init(). Tests for this and Calculator.#submitRequest() set up a console.error spy along with a callback for Vitest's vi.waitFor(). I need to write a document and/or blog post about this as part of the Handlebars Component Pattern. (I just came up with that name while writing it.)
mbland
added a commit
to mbland/test-page-opener
that referenced
this pull request
Jan 16, 2024
This commit solves the coverage drop from: - #23 01a79f6 Restores `istanbul` as the coverage reporter for the JSDom-based CI run to restore consistency in reporting, particularly in Coveralls. --- Adapted from the new test/missing-app-div/jsdom.test.js: Moved the "missing #app div" test into a separate directory from the other tests in ../jsdom.test.js. to prevent Node.js from using the same `test-modules/main.js` import. Otherwise: - If the test case were in the same file, the "missing #app div" branch wouldn't execute and the test would fail. - If this test file were in the same directory, the Istanbul coverage reporter wouldn't see the coverage from the "missing #app div" branch. I don't know exactly why that is. At the same time, the previous `src="./main.js?version=missing"` query suffix is no longer necessary. I got the idea to organize the tests this way after successfully covering similar code in: - mbland/tomcat-servlet-testing-example#85 mbland/tomcat-servlet-testing-example@b5df30e
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Adds Typescript to devDependencies and a jsconfig.json files to ensure Visual Studio Code performs more thorough type checking.
Description largely copied from:
The code is still JavaScript, but now we get strict type checking in Visual Studio Code and in continuous integration via
tsc
inpnpm typecheck
.The docs generated by 'jsdoc' are a little funky, and we don't get as much documentation in Visual Studio Code as I expected. I believe I can fix these issues at some point with this foundation in place.
The actual changes include:
Added @types/chai, jsdoc, and typescript as devDependencies.
Set .eslintrc to disable the no-undefined-types rule by extending "plugin:jsdoc/recommended-typescript-flavor-error". This is because the Handlebars types in lib/parser.js weren't trivial to replicate, and TypeScript finds those types just fine. This was based on advice from:
And:
More background:
no-undefined-types
additional files or have it auto-detect them (and finding unused types within the same set of files); avoid defects of tampering withno-unused-vars
gajus/eslint-plugin-jsdoc#99Added
settings.jsdoc.preferredTypes.Object = "object"
to .eslintrc to enable "Object.<..., ...>" syntax in a JSDoc@typedef
. Got rid of some extra whitespaces in .eslintrc, too.Added '.js' extension to all internal imports and added JSDoc comments everywhere reqired by
pnpm typecheck
.Added 'jsdoc-plugin-typescript' to the build to handle the TypeScript
import().Type
directives. This ended up pulling in the 'es-abstract' module, which blew up the pnpm-lock.yaml file. If I get an itch, I'll implement my own plugin one day and replace it.Updated
pnpm test-ci
to incorporatepnpm jsdoc
andpnpm typecheck
. Added 'jsdoc' to devDependencies to enable this.Added
null
checks everywhere reqired bypnpm typecheck
. Added tests to cover all thenull
cases.Added globals.d.ts and a
/* global STRCALC_BACKEND */
ESLint comment to calculators.js to properly type checkglobalThis.STRCALC_BACKEND
. Ironically, this required just referencing it asSTRCALC_BACKEND
withoutglobalThis
.Added a temporary components/template.d.ts containing the Handlebars Template() type declaration. Once I properly export those types from rollup-plugin-handlebars-precompiler, I'll remove it. (That plugin currently contains lib/template.d.ts, not types/template.d.ts.)
Added a test for the
#missing app element
case in main.js by adding a new test/main-missing-app-div.test.js. I need to port it to mbland/test-page-opener to solve the coverage problem encountered in:Updated
setupFetchStub()
to detect the type of thebody
argument and callJSON.stringify()
itself if it's anobject
. This eliminated the need for most callers to callJSON.stringify()
.Updated
StringCalculatorPage
with typing information and made it so that an empty object will stand in fornull
elements. This is playing loose with typing a bit, as anynull
s will cause errors showing unknown property access. But that seemed better than burdening all callers to do their ownnull
checks or workarounds.Of special note:
Added the
instantiate()
parameter to Calculator.init() to inject a Handlebars Template() function. This enabled testing that a missing#numbers
element was logged by Calculator.init().Tests for this and Calculator.#submitRequest() set up a console.error spy along with a callback for Vitest's vi.waitFor().
I need to write a document and/or blog post about this as part of the Handlebars Component Pattern. (I just came up with that name while writing it.)