Tests should give a helpful error message if grammar repos are not present about tree-sitter

[maxbrunsfeld]

@gnprice Did you run script/fetch-fixtures first? That script needs to be run one time before running the tests; it just clones some of the grammars such as tree-sitter-javascript and tree-sitter-python to use for integration testing. I should probably update these tests to fail with a better error message if those grammars are not present.

[gnprice]

Ah indeed, thanks! After running that script, script/test now runs through its tests without crashing. (It does give a couple of failures; I'll look around a bit and file a separate issue if that isn't already known.)

I think it would be very helpful to have the test check for what it needs and give a proper error message -- when it segfaults, that looks like "this doesn't work on my platform", or "this has a bug that gets tickled on my machine", not "I'm missing a setup step".

Even if sometimes it has to say something like "I'm not sure quite what happened, but there was an error (and here's one thing that sometimes causes that)", that would be much better than dereferencing NULL.

[gnprice]

(It does give a couple of failures; I'll look around a bit and file a separate issue if that isn't already known.)

Looks like they're determined by the random seed. Filed as #112!

[maxbrunsfeld]

I think it would be very helpful to have the test check for what it needs and give a proper error message.

I agree. I updated the title of this PR to capture that problem.

Filed as #112!

Ah yeah, I've noticed a few times that there's still have a bug in how we calculate which text ranges have changed their syntax after an edit. Thanks for taking the time to file an issue.

For context, I'm currently working on integrating Tree-sitter into Atom, and there is a bit of work to do on the Atom side, but once I'm done with that, I will fix this and a few other remaining blockers on the Tree-sitter side. It might be a couple of weeks before I get to it.

[gnprice]

For context, I'm currently working on integrating Tree-sitter into Atom, and there is a bit of work to do on the Atom side, but once I'm done with that, I will fix this and a few other remaining blockers on the Tree-sitter side. It might be a couple of weeks before I get to it.

Perfectly reasonable -- and thanks for the context!

[maxbrunsfeld]

Now they at least don't crash, and they report a somewhat suggestive error message (No such file or directory):

test tests::corpus_test::test_real_language_corpus_files ... 

Random seed: 1549479004

thread 'tests::corpus_test::test_real_language_corpus_files' panicked at 'called `Result::unwrap()` on an `Err` value: Os { code: 2, kind: NotFound, message: "No such file or directory" }', src/libcore/result.rs:1009:5