feat: migrate package to TypeScript and publish types #534

michaelfaith · 2025-07-27T19:11:49Z

This change migrates all source (runtime code and tests) to TypeScript, and as a result now results in type definitions being published with the package.

Because I knew this was going to be a big PR, I tried to break things up the best I could by commit. I first git mv all runtime files, and then in separate commits, migrated each rule. I then did the same thing for tests.

The actual changes to test files, beyond the file name change, was fairly minimal. Most of them were just simple file name changes. Only a few needed some additional work (e.g. types added to utility functions). So, I feel pretty good about maintaining functionality.

I will say, the type in utils aren't the best, and I had to resort to adding @ts-expect-error in a couple places, but even then, they're in a pretty decent state. I wanted to keep the flow of the functions as close to their original form as possible in this initial cut over. All of that is to say, there's opportunity to iterate on these internal types, to improve them further. Of course, those changes wouldn't be breaking and could potentially come after v7 is released.

In addition to all the runtime and test file changes, I addded a build script (using tsup) and updated the package.json with new entry point paths and to include dist instead of lib. I also updated the publish GH workflow to run a build before npm publish.

Supporting work that I landed prior to this:

migrate eslint.config.js to typescript so we could use the plugin directly from source: build: migrate eslint.config.js to typescript #531
Remove unneeded context compatibility functions: refactor: remove context compat functions #532
migrate the eslint-remote-tester config to be typescript so we could use the plugin directly from source: build: convert eslint-remote-tester config to typescript #533

Closes #310

migrate fixer-return migrate fixer-return

migrate consistent-output

bmish · 2025-08-02T14:49:58Z

lib/index.ts

@@ -41,20 +39,56 @@ import requireMetaType from './rules/require-meta-type.js';
 import testCasePropertyOrdering from './rules/test-case-property-ordering.js';
 import testCaseShorthandStrings from './rules/test-case-shorthand-strings.js';

+const require = createRequire(import.meta.url);
+
+const packageMetadata = require("../package.json") as {


I had issues with this which I fixed in qunitjs/eslint-plugin-qunit#584

That doesn't really apply in this case, since this file's relative position to the package.json is the same in the dist folder as it is in the lib folder.

Nice! Looks like I should be able to move index.js to fix that in my project:

refactor: remove workaround for package.json compiled path in index.js qunitjs/eslint-plugin-qunit#619

michaelfaith · 2025-08-02T20:56:27Z

I believe I've addressed all of the feedback so far. Let me know if there are any other updates you think should be made. Appreciate the thorough review.

michaelfaith · 2025-08-03T00:59:10Z

Added a typecheck step to the CI workflow

bmish

It's looking good. Just a few more fixes.

Would you also mind just running some manual sanity checks on the compiled output? E.g. try loading the plugin from dist/.

And make sure dist/ is ignored by tools like the following PRs (issues I discovered during the build/publishing process):

qunitjs/eslint-plugin-qunit#585 - eslint
qunitjs/eslint-plugin-qunit#602 - test coverage

bmish · 2025-08-03T15:31:54Z

.github/workflows/main.yml

+      - uses: actions/setup-node@v4
+        with:
+          node-version: 'lts/*'
+      - run: npm install


Technically, I believe we should be using npm ci in these workflows. Fine to consider it as a follow-up.

Yeah, I just matched the other steps. Could be a follow-up that changes them all to ci?

I was thinking a bit more on this later. In order to use npm ci, it'll require than just updating the calls in the workflow. The repo would need to start maintaining a lock file. It currently has lock files explicitly disabled, and npm ci requires that. I personally don't think having a lock file is a bad thing. I was kind of surprised when I first realized there wasn't one. But it seemed to be a conscious choice someone made at some point.

The other consideration is that using npm ci in the builds will result in longer build times, if that's important to you.

Oh good catch. I now remember that's why we don't use npm ci. Let's avoid adding it at this time. For us, it's fine to be more aggressive with dependency updates and omit a lockfile.

bmish · 2025-08-03T16:32:23Z

lib/index.ts

@@ -41,20 +39,56 @@ import requireMetaType from './rules/require-meta-type.js';
 import testCasePropertyOrdering from './rules/test-case-property-ordering.js';
 import testCaseShorthandStrings from './rules/test-case-shorthand-strings.js';

+const require = createRequire(import.meta.url);
+
+const packageMetadata = require("../package.json") as {


Nice! Looks like I should be able to move index.js to fix that in my project:

refactor: remove workaround for package.json compiled path in index.js qunitjs/eslint-plugin-qunit#619

lib/rules/require-meta-docs-recommended.ts

lib/utils.ts

bmish · 2025-08-03T16:54:22Z

package.json

@@ -75,6 +81,7 @@
    "npm-run-all2": "^7.0.1",
    "prettier": "^3.4.1",
    "release-it": "^17.2.0",
+    "tsup": "^8.5.0",
    "typescript": "^5.8.3",


In this or a later PR, we can upgrade this:

Suggested change

"typescript": "^5.8.3",

"typescript": "^5.9.2",

The version of @typescript-eslint that's installed doesn't include 5.9 in the peer deps yet. I'd suggest having a separate change that updates both sets of packages. I don't mind doing that one too.

bmish · 2025-08-03T16:56:01Z

tests/lib/utils.ts

+          return (
+            (
+              (
+                (
+                  (
+                    (
+                      (ast.body[0] as ExpressionStatement)
+                        .expression as AssignmentExpression
+                    ).right as ObjectExpression
+                  ).properties[0] as Property
+                ).value as ObjectExpression
+              ).properties[0] as Property
+            ).value as ObjectExpression
+          ).properties as Property[];


Just wondering what's going on here?

That's constructing the AST types that the tests all assume, based on the text of the string. It was extremely tedious, using AST Explorer and the test strings to figure it out 😅. There's not a great way to infer the types of what a string's AST will ultimately be parsed into, yet the tests all assume the structure of the string of the test case. Definite opportunity for a large refactor of how these tests operate.

michaelfaith · 2025-08-03T21:16:06Z

Would you also mind just running some manual sanity checks on the compiled output? E.g. try loading the plugin from dist/.

👍 Done. I ran a build and updated our repo's eslint.config.ts to use the package name instead of path the source in lib. Worked, as expected.

💡 Idea for follow-up: add an examples directory with different project configurations, that all have the package installed from the root of the repo. And then set up an end-to-end CI step that runs a build and then runs lint on each dir in examples. I don't really think that would be in the scope of this PR, but could be a nice follow-up task.

And make sure dist/ is ignored by tools like the following PRs (issues I discovered during the build/publishing process):

I already added dist to eslint.config.ts and .gitignore. I don't think there are any other configs that would need it? The test and coverage includes are already targeting lib/ and tests/, so those are fine.

Checking on the eslint.config.ts made me realize I still had the ts-expect-error on the plugin import, since that was set up when it was still js-only. I've removed that now.

bmish

Awesome work, thank you!

* main: feat: migrate package to TypeScript and publish types (eslint-community#534) build: convert eslint-remote-tester config to typescript (eslint-community#533) refactor: remove context compat functions (eslint-community#532) build: migrate eslint.config.js to typescript (eslint-community#531)

michaelfaith added 30 commits July 25, 2025 17:15

change test-setup to ts

2838607

add build config

8582fab

add types packages

046e58d

rename files to ts

d554cd3

migrate fixer-return

bc56d00

migrate fixer-return migrate fixer-return

migrate consistent-output

7e346f5

migrate consistent-output

migrate meta-property-ordering

98cf8a3

migrate no-deprecated-context-methods

ce6e93b

migrate no-deprecated-report-api

baedd5b

migrate no-identical-tests

5cee9cd

migrate no-meta-replaced-by

cb1d41d

migrate no-meta-schema-default

562fa66

migrate no-missing-message-ids

9c5e68e

migrate no-missing-placeholders

3694a1e

migrate no-only-tests

02314b4

migrate no-property-in-node

d86155c

migrate no-unused-message-ids

326cfce

migrate no-unused-placeholders

dbac215

migrate no-useless-token-range

912ab48

migrate prefer-message-ids

16430e0

migrate prefer-object-rule

c3b8257

migrate prefer-output-null

6080c2a

migrate prefer-placeholders

dbab18b

migrate prefer-replace-text

4a187f5

migrate report-message-format

10b1135

migrate require-meta-default-options

80b563c

migrate require-meta-docs-description

900ae8c

migrate prefer-replace-text

0eb77cc

migrate require-meta-docs-recommended

11fdcbf

migrate require-meta-docs-url

1ce2806

michaelfaith force-pushed the build/typescript branch from 27de051 to 64429ad Compare July 30, 2025 22:33

michaelfaith added 3 commits July 30, 2025 18:01

removed casting from no-meta-schema-default

3768cfa

remove casts from no-useless-token-range

2f80268

remove cast from report-message-format

1abe593

bmish added the enhancement label Aug 1, 2025

michaelfaith added 4 commits August 1, 2025 03:50

remove cast from require-meta-default-options

22cf264

remove cast from require-meta-docs-url

88bc975

remove casts from require-meta-fixables

9f0af6f

remove cast from require-meta-type

fb85ff6

michaelfaith force-pushed the build/typescript branch from 5d61735 to fb85ff6 Compare August 1, 2025 23:11

Adjust PartialRuleInfo types

1f22290

bmish reviewed Aug 2, 2025

View reviewed changes

addressed feedback in utils

6890444

michaelfaith requested a review from bmish August 2, 2025 20:56

ci: add typecheck step to CI workflow

ce4f96f

bmish requested changes Aug 3, 2025

View reviewed changes

removed unneeded ts-expect-error from eslint.config.ts

11fe179

michaelfaith added 2 commits August 3, 2025 16:23

Address feedback in utils

f287566

remove non-null assertion from require-meta-docs-recommended

9eeef96

michaelfaith requested a review from bmish August 4, 2025 00:39

bmish approved these changes Aug 4, 2025

View reviewed changes

bmish merged commit 95b859a into eslint-community:main Aug 4, 2025
6 checks passed

github-actions bot mentioned this pull request Aug 4, 2025

chore(main): release 7.0.0 #526

Merged

bmish mentioned this pull request Aug 4, 2025

V7 Planned Changes #505

Closed

michaelfaith deleted the build/typescript branch August 4, 2025 08:37

bmish mentioned this pull request Aug 4, 2025

chore: switch to TypeScript for .eslint-doc-generatorrc.ts #536

Merged

Uh oh!

feat: migrate package to TypeScript and publish types #534

feat: migrate package to TypeScript and publish types #534

Uh oh!

Conversation

michaelfaith commented Jul 27, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

michaelfaith commented Aug 2, 2025

Uh oh!

michaelfaith commented Aug 3, 2025

Uh oh!

bmish left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

michaelfaith commented Aug 3, 2025

Uh oh!

bmish left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

michaelfaith commented Jul 27, 2025 •

edited

Loading

bmish left a comment •

edited

Loading