An ESLint formatter that can report errors as todos, which can be deferred and fixed at a later time.
Linting is a fundamental tool to help ensure the quality of a codebase. Ensuring there are as few linting errors as possible (ideally 0), is a useful measure of a baseline of code hygiene.
It's common to leverage linting not just for syntax adherence, but also to direct developers to employ standardized patterns. As such, it's a fairly routine activity to introduce new lint rules into a codebase. This introduction, while necessary, can cause unintended friction, such as:
- new lint errors being introduced where they previously didn't exist
- causing unintended delays to shipping new fixes and features
- an "all or nothing" approach, where new rules require fixing before rollout
Having the ability to identify violations as todo
s allows for this incremental roll out, while providing tools that allow maintainers to view the full list of todo violations.
This formatter allows you to introduce new rules immediately, without blocking commits, by taking a snapshot of errors and transforming them to todos. Errors not found in this snapshot will continue to reported as errors.
Todos are stored in a .lint-todo
directory that should be checked in with other source code. Each error generates a unique file, allowing for multiple errors within a single file to be resolved individually with minimal conflicts.
To convert errors to todos, you can use the UPDATE_TODO
environment variable. This will convert all active errors to todos, hiding them from the linting output.
UPDATE_TODO=1 eslint --format @scalvert/eslint-formatter-todo
If you want to see todos as part of eslint
's output, you can include them
If an error is fixed manually, eslint
will let you know that there's an outstanding todo
file. You can remove this file by running --fix
eslint . --format @scalvert/eslint-formatter-todo --fix
Todos can be created with optional due dates. These due dates allow for todos to, over a period of time, 'decay' the severity to a warning and/or error after a certain date. This helps ensure that todos are created but not forgotten, and can allow for better managing incremental roll-outs of large-scale or slow-to-fix rules.
Due dates can be configured in one of two ways, but both specify integers for warn
and error
to signify the number of days from the todo
created date to decay the severity.
💡 Both warn
and error
are optional. The value for warn
should be greater than the value of error
.
- Via package.json configuration
{
"lintTodo": {
"decayDays": {
"warn": 5,
"error": 10
}
}
}
- Via environment variables
UPDATE_TODO='1' TODO_DAYS_TO_WARN="5" TODO_DAYS_TO_ERROR="10" eslint . --format @scalvert/eslint-formatter-todo
In order of precedence, environment variables override package.json configuration values.
For example, if you've specified the following values in the package.json configuration...
{
"lintTodo": {
"decayDays": {
"warn": 5,
"error": 10
}
}
}
...and you supply the following environment variables:
UPDATE_TODO='1' TODO_DAYS_TO_WARN= '2' eslint . --format @scalvert/eslint-formatter-todo
...the todos will be created with a warn
date 2 days from the created date, and an error
date 10 days from the created date.
Converting errors to todos with warn
and error
dates that transition the todo
to warn
after 10 days and error
after 20 days:
UPDATE_TODO='1' TODO_DAYS_TO_WARN= '10' TODO_DAYS_TO_ERROR='20' eslint . --format @scalvert/eslint-formatter-todo
Converting errors to todos with warn
and error
dates that transition the todo
error
after 20 days, but doesn't include a warn
date:
UPDATE_TODO='1' TODO_DAYS_TO_WARN= '' TODO_DAYS_TO_ERROR='20' eslint . --format @scalvert/eslint-formatter-todo