Fix bad_grouping_or_ordering_0 test

sinkingpoint commented 3 years ago

This test is really janky in that it actually tests 4 things, rather than just the metric interweaving it proports to test:

"A Histogram MetricPoint MUST contain at least one bucket (the given one doesn't)
"Histogram MetricPoints MUST have at least a bucket with an +Inf threshold" (the given one doesn't)
The sum without a count thing I mentioned in #229 (the histograms have sums without counts)
The metric interweaving (what it's meant to be testing)

This commit fixes the first three to make this only test one thing

brian-brazil commented 3 years ago

That would make sense, the test was designed to test the specific code of the Python parser. Can you fix the Go validator so it correctly rejects this?

brian-brazil commented 3 years ago

I think you forgot to push the changes to the Go code itself.

sinkingpoint commented 3 years ago

Sure. Think that commit should do it. It's not the cleanest, but it seems to work

sinkingpoint commented 3 years ago

Fixed those. Thanks for all your guidance @brian-brazil

brian-brazil commented 3 years ago

Thanks!

prometheus / OpenMetrics

Fix bad_grouping_or_ordering_0 test #232