#1818: implement --production flag

[aaron-j-krause]

Addresses #1818 and #1759.

Adds a --production flag that fails tests that are: pending, marked as skipped, marked as only, or have a duplicate title.

Since only is determined by grep the only way I could think to catch them is to fail a test if it sees a grep. Since this is, in a way, a linter meant to be run over all the tests before a commit or the like I figured that wouldn't be an issue. At the moment it only catches at the test level, so it won't catch suites marked as skipped or only.

[aaron-j-krause]

Any thoughts on this?

[danielstjules]

Thanks for the PR! :)

@boneskull what do you think?

[glenjamin]

My 2¢ - I'd want to fail on only or duplicate titles, but not on pending or skipped.

Only is something that should never be committed, if it's present then it must be a mistake (like leaving in a debugger statement).

Pending or skipped I'll sometimes use to allow merging code which will later need updates - I can configure CI tools to count these and show builds as unstable if needed.

[dasilvacontin]

My 2¢ - I'd want to fail on only or duplicate titles, but not on pending or skipped.

Same here.

Only is something that should never be committed, if it's present then it must be a mistake (like leaving in a debugger statement).

If I commit a .only or a debugger statement is for debugging purposes, or because it's a WIP commit – I want CI to fail the build.

I'm not fond of the flag's name, production, since that's not the condition under which I want .onlys to fail the build. I also want them to fail the build in PRs, in the dev branch, in the feature branch, in staging, etc. Probably everywhere but locally. So it would always make CI fail.

I wonder if there's anyone who doesn't want CI to fail when tests have .only, and why aren't they using pending tests instead. Maybe to make a CI build shorter when they are only interested on the output of certain test? dunno

[glenjamin]

I'm not fond of the flag's name, production, since that's not the condition under which I want .onlys to fail the build. I also want them to fail the build in PRs, in the dev branch, in the feature branch, etc. Probably everywhere but locally. So it would always make CI fail.

Perhaps --strict ?

[aaron-j-krause]

IIRC 'strict' was the original idea for the flag but 'production' was suggested in order to avoid similarities to 'strict mode' which it isn't really related to. Taking pending and skip filtering out of this PR would be easy enough. I could see the case for leaving pending in. Not so sure about skips, though.

[dasilvacontin]

IMO skip is for temporally (days, weeks, hopefully not more) disabling some tests, because they are no longer correct, they have to be reworked, the code they are covering is undergoing changes and I don't care about it being covered or not, whatever. So I want those to be committed, and not to break the build.

Sadly, I still don't have a better naming proposal, but I'll continue thinking about it.

On 15 Apr 2016, at 23:41, aaroncrows [email protected] wrote:

IIRC 'strict' was the original idea for the flag but 'production' was suggested in order to avoid similarities to 'strict mode' which it isn't really related to. Taking pending and skip filtering out of this PR would be easy enough. I could see the case for leaving pending in. Not so sure about skips, though.

—
You are receiving this because you commented.
Reply to this email directly or view it on GitHub

[boneskull]

👎 @aaroncrows is this suitable for the use case?

[boneskull]

A feature like this may be helpful, but since there's a (arguably better) workaround via ESLint, I'm not in a hurry to merge this...

[boneskull]

it doesn't cover skip (I think), but could be trivially extended to do so

[hollomancer]

This can be closed, due to changes in 3.0.0 in how the internals of it work. The original issue of #1818 needs a new PR.