Why does effective_dart cancel out the hints from pedantic?

[andesappal]

I had 87 hints from pedantic. Then I added effective_dart, and the hints disappeared.

[andesappal]

Also, could your team just automate the correction of pedantic hints? Or display an option to correct all hints?

[andesappal]

This is what I have:

I also noticed I have a warning "Key 'include' is duplicated" in

Before I saw this warning, was using pedantic 1.9.0 and effective 1.2.1

[bwilkerson]

We don't currently support including more than one other file in an analysis options file (this is the first use case I've heard of for it). There are some interesting questions about how the files would interact in case of conflicts that we decided to solve by disallowing that situation to arise.

My guess is that the second include replaces the first.

[andesappal]

Disallowing it doesn't solve it, right? But that's ok, I'll use one analysis option at a time. Thank you.

[andesappal]

Maybe what's effective implies new code, thereby invalidating the current pedantic hints. I see, it does pose an interesting problem.

[bwilkerson]

The problems related to multiple includes are around conflicts. If one says to enable a lint and another says to disable it, which one wins? With a single include those problems don't exist.

And we defined the "conflicts" between a file and the file it includes to be resolved by saying that the file containing the include always takes precedence.

[domesticmouse]

Can we get this re-opened? Being able to depend on both of the pedantic and effective_dart linters is really needed. Last one wins, potentially with a warning. Or even make it an error.

[bwilkerson]

I've re-opened the issue, but I'm not sure what you're asking for. We produce a diagnostic (with a severity of error) telling the user that we don't support having multiple includes (or, more technically, the YAML parser tells the user that duplicate keys are not allowed).

Being able to depend on both of the pedantic and effective_dart linters is really needed.

I'd be interested to understand more about why this is needed and just how badly the lack of support is impacting our users. I'm not opposed to adding support, but we need to prioritize our efforts.

Unfortunately, supporting the inclusion of multiple analysis options files is a non-trivial effort. If we only needed to worry about lints, then I suspect that the merge semantics would be fairly easy to define. But we're including the whole analysis options file, and the semantics of merging all of the options isn't quite as straight-forward. (We actually spent some time when first adding inclusion support attempting to do just that and decided that it was more important to be able to have a single chain of inclusion than to postpone the release while we figured it all out.)

[domesticmouse]

The core issue is that the two list of lints contain disjoint recommended lints. I want to be able to recommend a best practices setting that contains both.

One work around would be to merge the two sets of lints into a best practices set and publish that list.

[bwilkerson]

@mit-mit

[mit-mit]

@domesticmouse can you elaborate on why this is really needed? I'm not sure I understand the use case.

[domesticmouse]

Let me turn the question around the other way. Why are there lints included in the effective_dart configuration, which AFAICT enforce the best practices we document at dart.dev/guides/language/effective-dart, but are not int the pedantic configuration?

[bwilkerson]

The pedantic package represents the collection of lints that are enabled for our internal users. The internal community has not chosen to adopt all of the lints in effective_dart for various reasons. Sometimes it's because of the number of false positives, sometimes for other reasons.

[domesticmouse]

False positives are bad. I'll stick with just recomending pedantic then.