Minimum requirement

[InEdited]

Added an option to put a minimum coverage constraint.
This resolves issue #68

• By default minimum coverage constraint is zero
• Number in config file represents the required minimum percentage of total coverage
• Process exits by an error but still outputs the test report

[InEdited]

Just noticed that using this feature with the one implemented in either pull requests #72 or #71 will cause the reporting to always fail, will work on that to make it work only on the tested files.

[hishamhm]

Just noticed that using this feature with the one implemented in either pull requests #72 or #71 will cause the reporting to always fail, will work on that to make it work only on the tested files.

#71 is now merged, could you adapt this PR to be based on the current master code instead?

Also, I think users would expect the minimum requirement % calculation to include the untested files when includeuntestedfiles is set.

[allbarbos]

Can I open another PR with the change?

[esatterwhite]

Is this still in progress. Would love to get this feature landed

[hishamhm]

@allbarbos Yes, please!

(I'm very low on bandwidth for maintaining LuaCov — in fact, the previous maintainer passed away and I'm carrying the torch, but I'd love to get some maintainership help on this! If any of you wants to help out, please update this PR and let me know and I'll grant commit access)

[Tieske]

@hishamhm I fixed up the PR. But it has issues beyond the simple feature;

• the message and error value is generated in the default-reporter. Using another reporter will not behave the same, depending on how it was created.
• moving it into the generic machinery is tricky, since you do not want an early exit because of this if the report has to be uploaded a coverage tool (eg. coveralls)

A simple option would be to leave this as it is today, and have a CI user run 2 reports. First the one they want for display/reporting. And then the default one, which will display the error.

(though that means one cannot configure the stats files to be cleared, since the second run would need those as well)

[Tieske]

I'm leaning towards a separate reporter that doesn't really report anything, but just generates the error message.