Update

[jugglinmike]

Hi @bakkot,

We elected to change the specification on the interpretation of function names this month, and I've recently submitted a spec patch to do just that. I wanted to follow up by reflecting that decision in this project.

It turns out that we had coverage for the old expected behavior. This made my job easy: I just moved the files. The problem is that these tests originate from the Acorn project. I thought this would mean that Acorn also needed to be updated (in contrast to my prior research), but it turns out that the relevant tests in Acorn already classify these tests as valid syntax. It seems likely that they were intentionally re-classified in order to match the specification at the time they were imported, but I wanted to draw your attention to this in order to be sure (and to explain why no change in Acorn seems to be necessary).

(I've included the script I used to research this below.)

I've also added a few more tests to improve coverage. Have you accepted novel test material in this project? Or should I be landing the new tests in one of the upstream projects first?

Commit message:

Relocate tests for BindingIdentifier of strict fns

The modified test material was introduced via commit 73c001c on 2017-04-26. According to this project's licenses.md file, it originated from the Acorn project. At that that date, the tests containing the relevant productions were labeled as negative syntax tests. However, the tests were introduced to this project as positive syntax tests.

Although the new interpretation was technically correct in terms of the ECMA262 specificatoin at that time, TC39 reached consensus in March 2018 to classify these productions as syntax errors.

Re-locate the relevant tests to align with the decision. Note that due to the discrepency described above, no change to the source project is necessary.

Research script:

#!/bin/bash

test_files='
pass/050a006ae573e260.js
pass/2c0f785914da9d0b.js
pass/574ea84fc61bdc31.js
pass/6c4fe38464c16309.js
pass/8643da76fe7e95c7.js
pass/e0c3d30b6fe96812.js'

for test_file in $test_files; do
  echo $test_file:
  git log --format='%h %cd %s' -- $test_file
done

grep -E -e $(echo $test_files | sed 's/ /|/g') -e '^##' licenses.md

(
  cd ../acorn
  git checkout $(git log -n 1 --format='%h' --since 2017-04-26 master)
  grep -E 'function\s+static' test/tests.js
)

And the output:

pass/050a006ae573e260.js:
73c001c Wed Apr 26 15:41:55 2017 -0700 Switch names to slugs
pass/2c0f785914da9d0b.js:
73c001c Wed Apr 26 15:41:55 2017 -0700 Switch names to slugs
pass/574ea84fc61bdc31.js:
73c001c Wed Apr 26 15:41:55 2017 -0700 Switch names to slugs
pass/6c4fe38464c16309.js:
73c001c Wed Apr 26 15:41:55 2017 -0700 Switch names to slugs
pass/8643da76fe7e95c7.js:
73c001c Wed Apr 26 15:41:55 2017 -0700 Switch names to slugs
pass/e0c3d30b6fe96812.js:
73c001c Wed Apr 26 15:41:55 2017 -0700 Switch names to slugs
## Acorn
* pass/050a006ae573e260.js
* pass/2c0f785914da9d0b.js
* pass/574ea84fc61bdc31.js
* pass/6c4fe38464c16309.js
* pass/8643da76fe7e95c7.js
* pass/e0c3d30b6fe96812.js
## Esmangle
## Esprima
## Shift
## UglifyJS2
testFail("function static() { \"use strict\"; }",
testFail("\"use strict\"; function static() { }",

[bakkot]

Thanks for the patch!

It seems likely that they were intentionally re-classified in order to match the specification at the time they were imported

Yup. I used upstream tests as a source, but did my own classification.

Have you accepted novel test material in this project?

No, but we can start doing so. I'll push a commit to your branch updating the licenses file (which needs to be updated in light of these tests being moved anyway).

[bakkot]

Thanks!

[jugglinmike]

My pleasure!