fix: show correct stack trace in errors and console

[sheremet-va]

Fixes #2231
Fixes #1331
Fixes #2211
Fixes #2195

We might want to update to @cspotcode/source-map-support, but currently it doesn't work with fastify: cspotcode/node-source-map-support#44

[sheremet-va]

@haikyuu hey, tests with imba are failing with this PR. I checked source maps with https://sokra.github.io/source-map-visualization/#custom and it seems they are wrong? I wonder how it worked before 🤔 Can you help me with figuring this out?

[haikyuu]

@sheremet-va let me look at this tomorrow (it's night for me now). Fun thing is that I was planning on fixing source maps in Imba as well.

[haikyuu]

@sheremet-va the sourcemaps in the test seem to be correct here

[sheremet-va]

@sheremet-va the sourcemaps in the test seem to be correct here

This is not what is actually running tho. This is the final code:

/*body*/
const __vite_ssr_import_0__ = await __vite_ssr_import__("/@fs/<path>/Projects/vitest/packages/vitest/dist/index.js");


function add(...args){

        return args.reduce(function(a,b) { return a + b; },0);
}
Object.defineProperty(__vite_ssr_exports__, "add", { enumerable: true, configurable: true, get(){ return add }});;

__vite_ssr_import_0__.it("add",function() {

        __vite_ssr_import_0__.expect(add()).toBe(0);
        __vite_ssr_import_0__.expect(add(1)).toBe(3);
        return __vite_ssr_import_0__.expect(add(1,2,3)).toBe(6);
});

For some reason source map is not updated, when transformed with SSR 🤔 @antfu do you know what can be the cause for this?

[sheremet-va]

Hm, it seems imba files return inlined maps in transformRequest 🤔 So we don't inline updated source map, and use theirs

@haikyuu is there any reason why you inline source maps instead of relying on Vite support?

[haikyuu]

@sheremet-va let me check

[sheremet-va]

@sheremet-va let me check

So, I checked other plugins (Vue, for example), and I see they return source map in Rollup convention way inside map property in transform hook, so other tools/plugins can reuse them. I would expect imba to do the same in this case 🤔

I would appose replacing source provided inlined maps, because we might replace our own source maps several times, which might case performance issues.

If I remove inlined source maps from a plugin, tests pass.

[haikyuu]

@sheremet-va I tested providing external source maps in the imba plugin. And it works correctly now 💯

[rcoundon]

We're still seeing problems with line numbers with version 0.25.3. When there's a failure and the line numbers in the stack trace are shown in the console they're way out. Also, when debugging (using both WebStorm and VSC) the debugger doesn't track the executing line.

Is there anything specific we need to do fix this in terms of configuration or do you think there could still be an underlying issue?

[sheremet-va]

We're still seeing problems with line numbers with version 0.25.3. When there's a failure and the line numbers in the stack trace are shown in the console they're way out. Also, when debugging (using both WebStorm and VSC) the debugger doesn't track the executing line.

Is there anything specific we need to do fix this in terms of configuration or do you think there could still be an underlying issue?

Please, just provide reproduction. Small talk doesn’t help fixing the issue. Yes, you might have the problem. If you can give me a small reproducible example, I can check if there is an error and why.

[rcoundon]

We're still seeing problems with line numbers with version 0.25.3. When there's a failure and the line numbers in the stack trace are shown in the console they're way out. Also, when debugging (using both WebStorm and VSC) the debugger doesn't track the executing line.
Is there anything specific we need to do fix this in terms of configuration or do you think there could still be an underlying issue?

Please, just provide reproduction. Small talk doesn’t help fixing the issue. Yes, you might have the problem. If you can give me a small reproducible example, I can check if there is an error and why.

Apologies, but that wasn't meant as small talk, I asked a specific question about whether there's anything specific we need to do to make this work. I was also curious if this was supposed to be fixed, and if others were experiencing anything similar since this pull request is showing as merged but not closed.
I'll try to create a reproduction in the meantime.

[rcoundon]

Also, I'm trying to make sense of all the disparate information about this and related issues, for example, this thread and to be honest, I'm a bit lost as to what should and shouldn't work

[sheremet-va]

Also, I'm trying to make sense of all the disparate information about this and related issues, for example, this thread and to be honest, I'm a bit lost as to what should and shouldn't work

Source maps should just work.

[rcoundon]

In this example, the line numbers in the failed tests that are reported are correct, However, step debugging in WebStorm has the selected line jumping about in the file.
In two other projects the reported line numbers are incorrect but I haven't yet managed to isolate that in a small reproduction but I'll keep trying

[rcoundon]

Possibly related to Webstorm rather than vitest - https://youtrack.jetbrains.com/issue/WEB-54437#focus=Comments-27-6696752.0-0

[sheremet-va]

Possibly related to Webstorm rather than vitest - https://youtrack.jetbrains.com/issue/WEB-54437#focus=Comments-27-6696752.0-0

This PR might fix it: #2433

[rcoundon]

I've just tried with v0.25.4 but unfortunately the situation is actually worse in Webstorm. The IDE doesn't hit a breakpoint now regardless of what is or isn't set in the esbuild property of vite.config.ts.

Previously, no esbuild setting at all or sourcemap set to true or 'external' worked.

[sheremet-va]

I've just tried with v0.25.4 but unfortunately the situation is actually worse in Webstorm. The IDE doesn't hit a breakpoint now regardless of what is or isn't set in the esbuild property of vite.config.ts.

Previously, no esbuild setting at all or sourcemap set to true or 'external' worked.

Works fine with provided reproduction 🤷🏻

#2427

[rcoundon]

I've just tried with v0.25.4 but unfortunately the situation is actually worse in Webstorm. The IDE doesn't hit a breakpoint now regardless of what is or isn't set in the esbuild property of vite.config.ts.
Previously, no esbuild setting at all or sourcemap set to true or 'external' worked.

Works fine with provided reproduction 🤷🏻

#2427

I'm afraid it doesn't for me. I've just pushed the update to vitest to here and the breakpoints aren't hit with sourcemap: 'both'

[sheremet-va]

I see that stack trace is correct 🤷🏻 :