-
Notifications
You must be signed in to change notification settings - Fork 285
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[WIP] New rule use-errors-new
(propose replacing fmt.Errorf
by errors.New
when possible)
#1136
Conversation
The rule name
There is a linter called I have done the replacements to
|
I was about say there might be another linter that may have the rule. But I'm fine with having it supported in revive, as it's a specific tool. But then later, golangci-lint team might report it as a duplicate, maybe they will simply disable it by default. BTW, do you plan to enable it my default? Or at least to add it to the recommended rules list? |
rule/use_errors_new.go
Outdated
"github.com/mgechev/revive/lint" | ||
) | ||
|
||
// UseErrorsNewRule lints given else constructs. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I don't get this sentence, maybe you could rephrase
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
done
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
No test?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
As I've said in the PR description, it is work in progress, I'll complete the PR (tests+doc) in the next days.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
tests and doc added
I checked go vet, go critic, errorlint, gosimple. It's not supported For me, this rule should be a staticcheck one, but each repository deserves a linter maybe. The main issues with different tools, is the fact you are never sure something will get merged by a maintainer. Here we are in revive, you decide. But once in golangci-lint, someone may raise a flag. What do you think @ldez? |
It already exists inside golangci-lint:
|
To be clear, I have no problem with duplicate rules between golangci-lint and revive. The 2 projects are independent, so it's expected to have duplicates especially with "metalinters". Same thing with golangci-lint and staticcheck, or staticcheck and revive, or gocritic and revive, etc. I try to limit them because it's confusing for users to have 2 elements for the same rule/check but it should not be used as a limitation for other tools. But I have a problem when someone uses a PR open on Golangci-lint, and so the work of someone else, to create a rule inside another tool. |
Yes, it does. @alexandear already spotted it, and I was looking at other linters (even the experimental ones in gocritic, go vet and gosimple) if it was already available. And yes, apparently, perfsprint is currently the only one to report it.
Thanks for your feedback @ldez perfsprint detects way more things than what this rule would do, so I would say it's OK, anyway. It's not as if perfsprint was detecting only what is added in this PR. And then, it could become a deprecation logic at some point. So now, if someone want to have revive It might become a problem only if Thanks again Ludovic.
Yes, this one was pretty crazy |
Thank you all for the feedback. I didn't know about
I'm not planning to enable it by default.
Users with
OK, I'll stick to it if I don't find something better. |
rule/use_errors_new.go
Outdated
Category: "errors", | ||
Node: n, | ||
Confidence: 1, | ||
Failure: "replace fmt.Errorf by errors.New ", |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
nit:
Failure: "replace fmt.Errorf by errors.New ", | |
Failure: "replace fmt.Errorf by errors.New", |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
fixed
config/config.go
Outdated
&rule.UseErrorsNewRule{}, | ||
&rule.RedundantBuildTagRule{}, | ||
&rule.UseErrorsNewRule{}, |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Twice?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Likely to be with conflicts, this might explain why #1142 was opened
} | ||
|
||
if len(funcCall.Args) > 1 { | ||
return w // the use of fmt.Errorf is legit |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Well, not exactly
We have crazy people out there
fmt.Errorf("%s, "whatever")
https://github.com/search?q=language%3Ago++%22fmt.Errorf%28%5C%22%25s%5C%22%2C+%5C%22%22&type=code
fmt.Errorf("%w", err)
https://github.com/search?q=language%3Ago++%22fmt.Errorf%28%5C%22%25w%5C%22%2C+%22&type=code
I created an issue on perfprint linter, as it's in their scope
I think you can merge like this, with current code.
My point can be addressed later
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Discussion moved on the PR
Moved to #1142 |
WIP PR implementing a new rule
use-errors-new
that proposes replacingfmt.Errorf
byerrors.New
when possible.As indicated in this issue, beyond style considerations,
errors.News
is ~4x faster thatfmt.Errorf
I've tested the rule against public code bases like terraform and gitea and it finds hundreds of "failures" (with no false positives, the rule is very simple)
I'm not so happy with the rule name, if you have something better... I take.
I'll complete the PR (tests+doc) in the next days, meanwhile comments are welcome.
cc: @ccoVeille @alexandear