feat: report errors in /try via @codemirror/lint

push-f.com commented

2026-04-21 18:14:41 +02:00

First-time contributor

No description provided.

push-f.com added 2 commits

2026-04-21 18:14:42 +02:00

refactor: use an importmap for /try b0bc2aa297

In the next commit we'll introduce lint diagnostics for errors in the try
editor. As a prepration we stop rewriting imports before execution in this
commit since that can mess up reported ranges.

feat: report errors in /try via @codemirror/lint d2f93c79e5

marijn commented

2026-04-22 11:19:30 +02:00

Owner

This breaks path-independence by hard-coding absolute URLs, meaning the sandbox no longer works under the /website path of the devserver (it tries to read http://localhost:8090/try/mods/codemirror.js, though that's being served ad /website/try/mods/...).

As a general pointer, it's a good idea to propose new features as an issue first, before creating a PR. I'm not fond of increasing the scope of the code I maintain, nor do I particularly enjoy reviewing code, so I'm not generally keen on PRs that add features that weren't even on my radar. If you can get the import map to work, that would be a clear improvement here, so I'm cautiously positive towards this one, but in other cases, you might end up wasting your time implementing something I'm not interested in.

This breaks path-independence by hard-coding absolute URLs, meaning the sandbox no longer works under the /website path of the devserver (it tries to read http://localhost:8090/try/mods/codemirror.js, though that's being served ad /website/try/mods/...). As a general pointer, it's a good idea to propose new features as an issue first, before creating a PR. I'm not fond of increasing the scope of the code I maintain, nor do I particularly enjoy reviewing code, so I'm not generally keen on PRs that add features that weren't even on my radar. If you can get the import map to work, that would be a clear improvement here, so I'm cautiously positive towards this one, but in other cases, you might end up wasting your time implementing something I'm not interested in.

push-f.com force-pushed try-lint from d2f93c79e5 to c62d5b9b6a

2026-04-22 11:50:03 +02:00

Compare

push-f.com commented

2026-04-22 11:55:46 +02:00

Author

First-time contributor

Pushed a fix for the path-dependence. Showing errors in the editor is such a clear improvement to me, that I assumed you'd be interested. The importmap change happened in drive by. Point taken though, I'll open issues for other features.

marijn commented

2026-04-23 08:06:49 +02:00

Owner

Since these errors come in asynchronously (via postMessage over an async channel, or even because the code that triggers them didn't run synchronously at the start of the script), the code may be edited between the point where it is evaluated and the point where the error position arrives in the main window. When that happens, the error will be shown in the wrong position.

Since these errors come in asynchronously (via `postMessage` over an async channel, or even because the code that triggers them didn't run synchronously at the start of the script), the code may be edited between the point where it is evaluated and the point where the error position arrives in the main window. When that happens, the error will be shown in the wrong position.

push-f.com force-pushed try-lint from c62d5b9b6a to 2b5c44e71f

2026-04-23 08:14:20 +02:00

Compare

push-f.com commented

2026-04-23 08:16:41 +02:00

Author

First-time contributor

Good catch. I added a check to only show them if the document hasn't changed.