1.12.1 performance regression.

[matter123]

line 3493 is meta.function.definition
line 4427 is :function_call
line 633 is :source_wrapper

Edit: possibly duplicate of #256
Edit: not fixed by #258

[jeff-hykin]

Crap what files are those showing up in?

[matter123]

It is one of my own files, Ill try to find a repro that is a more reasonable size.

[jeff-hykin]

if you're okay with it being open sourced, I dont think the size will be an issue if you want to add it to the testing examples

[matter123]

At least in the branch Fix/#260, npm run perf seems to show the issue. I'll create a PR for adding it to the examples in an hour or two.

[jeff-hykin]

After running perf I get something like

How do I know what file that belongs to?

[matter123]

Unfortunately, that information is not recorded.

[jeff-hykin]

Ah haha well thats problematic, is there a way to program it in? or is restricted by a library

[matter123]

It's certainly possible, but currently the pattern timing recording doesn't distinguish between files.

[jeff-hykin]

Okay I'll work on adding that since I'm not sure how else to debug the slowdown

[matter123]

My best bet for how to add it would be have commands/perf inform the recorders on file change. And have a separate bin of records per file.

[matter123]

PR has been submitted with the file I originally ran perf on.

[jeff-hykin]

I created a very jank solution npm run perfall which takes about 10 minutes to run and outputs the performance metrics to a report.json file in the root dir.

[matter123]

Fixed in v1.12.15.