Snowpack 2.8.0 doesn't support "alias" anymore

[jcschmidig]

After an update to 2.8.0 snowpack build does not recognize the alias defined in snowpack.config.json.
"alias": {
"%router": "./src/router",
"%user": "./src/store/user"
},

Could not resolve unkonwn import "%router"

[FredKSchott]

Hm, if you remove "%" from the import, does it still work? That feature is definitely meant to still exist, but the special character may be messing with our resolution algorithm.

[jcschmidig]

I changed "%router" to "router" and now am getting:

[snowpack] ✔ build complete [7.76s]
[snowpack] ! installing dependencies…
[snowpack] Package "router" not found. Have you installed it?
[snowpack] Install failed.

[jcschmidig]

That means that an alias with a leading "%" does trigger another error then an alias without.

[FredKSchott]

right. Just confirming: if you change both the config "%router" to "router", and the import statement "%router" to "router" in your code, you still see that package not found error?

[jcschmidig]

Yep.

[FredKSchott]

Hmm, okay I'm sorry but I can't reproduce. We have an entire test for this here: https://github.com/pikapkg/snowpack/blob/master/test/build/resolve-imports/package.json#L10-L14

If you can edit this test to reproduce the error that you're seeing, we can take a look!

[jcschmidig]

I took your test package and added

1. in package.json/snowpack/alias
   "@comp": "./src/components"

2. in src/index.js
   import components_______ from '@comp'; // bare import using short alias

This works! (with version 2.7.x of snowpack)

But then I changed in package.json/dependencies
"snowpack": "2.8.0"

and it produces

[snowpack] ! building source…
[snowpack] ✔ build complete [0.04s]
[snowpack] ! installing dependencies…
[snowpack] Package "@comp" not found. Have you installed it? 
[snowpack] Install failed.

I assume that since 2.8.0 the completion of short alias with index/index.js disappeared.

[FredKSchott]

I'm sorry, I still can't reproduce with that exact example, on master.

I'm about to release a new version, so there's a chance that might fix your issue and explain why i can't reproduce. But, if are still seeing this issue on the latest release that goes out later today, then I'd really appreciate if you could add the failing test as a PR in this repo, so that we can take a look and fix.

[jcschmidig]

Et voilà: PR 886.
I hope this helps!

[FredKSchott]

Ah, thank you! I'll take a look and see what we can do