suggestion: alternative check for multiple graphql packages

[phryneas]

This could in the long term enhance early detection if users bundle multiple versions of the graphql package, and (in a future version of graphql) retire the current check that needs inside of instanceOf and thus needs to be restricted to development node, since it slows down the function.

[netlify]

✅ Deploy Preview for compassionate-pike-271cb3 ready!

Name	Link
🔨 Latest commit	422f59a
🔍 Latest deploy log	https://app.netlify.com/sites/compassionate-pike-271cb3/deploys/6492f8e5d6d45a00075295c4
😎 Deploy Preview	https://deploy-preview-3915--compassionate-pike-271cb3.netlify.app
📱 Preview on mobile	Toggle QR Code... Use your smartphone camera to open QR code link.

---

To edit notification comments on pull requests, go to your Netlify site settings.

[linux-foundation-easycla]

The committers listed above are authorized under a signed CLA.

• ✅ login: phryneas / name: Lenz Weber-Tronic (422f59a)

[yaacovCR]

This is really interesting! Is there the possibility that this side effect might get pruned by tree shaking? Might we need to list it formally as a side effect?

Note also that the current check emits a different instanceof function entirely so it also runs only once.

[phryneas]

@yaacovCR something like this shouldn't get tree-shaken, as it's not a declaration, but a method-call, which bundlers will assume as "with side-effects", unless you add a /** PURE */ annotation.

And even if it gets tree-shaken: nothing will break, it will just not do the check.

That said, this approach has the drawback that it will only work if all graphql versions that take part use it. If we'd ship this in v17, but not v15 and v16, it would only warn on different v17 versions, but wouldn't trigger with v16 and v17 in parallel.

So: not sure if it really makes sense - on the other hand, if we never start shipping it, it will never start making sense.

[yaacovCR]

Does it matter that we have sideEffects set to false?

Does it depend on the bundler?

evanw/esbuild#1241 (comment)

[phryneas]

sideEffects should stay false either way - while this is a side effect, it's not a very observable one or one that has an impact on anything than this package.

With sideEffects true, no tree-shaking will happen at all.

Beyond that, it is up to the bundler, but as far as I'm aware I haven't seen a bundler that has removed a top-level function call (that was not explicitly marked as pure) in a file that was imported for some other reasons.

The examples you linked to are not dropped because they are function calls (some bundlers mark console.log as pure, but most don't) that cannot be determined to be 100% pure.

Tree-shaking usually either drops full files (if none of the exports are used), or export definitions that are pure, either by annotation or by heuristics.