[savetheclocktower]

…on the command line.

Fixes #1399.

I explained the bug in this comment. This implements the described fix.

I was going to add a spec to cover this, but spec/main-process/atom-application.test.js seems to rely deeply on the idea that two subdirectories are created, and that all CLI tests will be invoked from their parent directory. Hence there's no way to test something like pulsar FOO.md:10 within this structure; only, e.g., pulsar a/FOO.md:10.

This could be fixed, but I think it's better to fix the underlying issue than to hold it up while I try to re-examine the assumptions of the test framework.

Manual testing

To ensure this is fixed, test the exact behavior that's failing on master:

• Invoke Pulsar from a directory with a root file, like a README.md, and include a line-number specification: pulsar README.md:10.
• Ensure that a window opens with one item: the README.md. The cursor should be on line 10.

[DeeDeeG]

Small change and I can confirm the fix using the steps to reproduce at the bottom of the PR text. 👍

The comment feels a little arcane without a ton of context, but the more I think about it the clearer it becomes... IDK if there's a simpler or more approachable phrasing, but also I don't think it's such a problem or important enough to block merging, so I think we're good!

[DeeDeeG]

Actually let me look into the editor tests for a bit, IDK if those failures are also on master? Looking into it for a sec.

[confused-Techie]

Can tell from the surrounding code that pathToOpen likely isn't the value we want to be checking here. Easy approval

[DeeDeeG]

Okay, we got a passing CI run on my placeholder commit (b3aefbb), ignoring the bugged duplicate package test job that is perma-pending (that package's tests did run in the original/non-duplicate job!).

I'll back that one out now and this should be good to go!