Fix ns export when between a default export and exported specifiers.

[jdalton]

Q	A
Bug fix?	yes
Breaking change?	no
New feature?	no
Deprecations?	no
Spec compliancy?	yes
Tests added/pass?	yes
Fixed tickets	127
License	MIT

A continuation of #127. I noticed that this will error:

export B, * as A, { C, D } from 'test';

The bit that's throwing it off is the last specifier , { C, D }.

Related to benjamn/reify#162.

[codecov]

Codecov Report

Merging #525 into master will increase coverage by <.01%.
The diff coverage is 100%.

@@            Coverage Diff             @@
##           master     #525      +/-   ##
==========================================
+ Coverage   98.27%   98.27%   +<.01%     
==========================================
  Files          22       22              
  Lines        3530     3531       +1     
  Branches      979      979              
==========================================
+ Hits         3469     3470       +1     
  Misses         22       22              
  Partials       39       39

Flag	Coverage Δ
#babel	81.05% <100%> (ø)	⬆️
#babylon	97.11% <100%> (ø)	⬆️

Impacted Files	Coverage Δ
src/parser/statement.js	99.08% <100%> (ø)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 4ef964e...d5a6a77. Read the comment docs.

[loganfsmyth]

My understanding is that this is expected behavior. These proposals essentially add export versions of the existing import forms, and the ES6 spec for import allows one, but not both of {foo as foo} or * as foo.

Note in the grammar: https://www.ecma-international.org/ecma-262/7.0/#sec-imports

ImportClause:
    ImportedDefaultBinding
    NameSpaceImport
    NamedImports
    ImportedDefaultBinding, NameSpaceImport
    ImportedDefaultBinding, NamedImports

[jdalton]

Thanks, I'll dig a bit and report back.

Update:

Ok, I've pinged Brian Terlson for insight into the design choice.
This can be closed though since it would be yet another proposal to remove binding order significance.

[jdalton]

Just pinging back I made reify support unordered export and import specifiers.
It simplified the export extension a bit too ⚡️

[loganfsmyth]

Sounds good. I don't think I'm comfortable landing any changes to this unless there were an official proposal, as you mentioned. I agree it's surprising behavior to me from a grammar definition standpoint, but being that this is the first time I've seen the question raised, I can't imagine it's a case that comes up often.

I think every time I've wanted a namespaced import and a named import, I've been fine to reference it off of the namedspace object instead. To me that says that this issue is more annoying for those writing tooling based around adding imports, since it makes adding imports more complicated, but it doesn't make it harder for day-to-day devs to get their work done.

[jdalton]

I don't think I'm comfortable landing any changes to this unless there were an official proposal, as you mentioned.

No worries 😎

I agree it's surprising behavior to me from a grammar definition standpoint, but being that this is the first time I've seen the question raised, I can't imagine it's a case that comes up often.

It's a bit WTFJS so I'm ok smoothing over its rough edge in my neck of the woods.