Migrating from ign-lint

iglint has an equivalent for every rule of ign-lint 0.6.4 and reads ign-lint's configuration (rule_config.json) and whitelist files. There are two ways to switch:

• Convert once with iglint migrate, and use iglint check from then on. Recommended.
• Keep the old command line with iglint ign-lint, which accepts ign-lint's arguments and prints ign-lint's output format.

Rule mapping

ign-lint rule	iglint
NamePatternRule	NAME001
PollingIntervalRule	BIND004
PylintScriptRule	SCR900 (the real pylint) and every default SCR rule that no other ign-lint rule maps to
BadComponentReferenceRule	VIEW018
ComponentReferenceValidationRule	BIND005 (expressions and property bindings) and SCR305 (scripts)
UnusedCustomPropertiesRule	VIEW015
ExcessiveContextDataRule	VIEW006
PropertyAccessRule	VIEW012
PropertyPersistenceRule	VIEW013

ign-lint's option names are mapped too, so a rule_config.json configures the iglint rules (minimum_interval becomes minimum-interval, and so on).

Option 1: iglint migrate

Run it next to rule_config.json. --dry-run prints the result without writing files:

iglint migrate --dry-run

For a rule_config.json that configures NamePatternRule and PollingIntervalRule and disables PylintScriptRule, the output starts like this (shortened):

# iglint configuration, migrated from rule_config.json by `iglint migrate`.

[rules.NAME001]
# name-pattern: Name does not follow the configured naming convention, from ign-lint NamePatternRule.
convention = "PascalCase"
target-node-types = ["component"]

[rules.BIND004]
# polling-interval: Expression polls faster than the minimum interval, from ign-lint PollingIntervalRule.
severity = "error"
minimum-interval = 5000

[rules.SCR001]
# syntax-error: Script does not parse with Jython 2.7, from ign-lint PylintScriptRule.
enabled = false

# ... one entry per native SCR rule, all disabled because PylintScriptRule is ...

# ign-lint runs every rule that is not disabled, on `view.json` files only.
[[overrides]]
files = ["**/view.json"]
select = ["NAME001", "VIEW012", "VIEW013", "VIEW018"]

followed by a summary of the mapping:

  NamePatternRule -> NAME001
  PollingIntervalRule -> BIND004
  PylintScriptRule: disabled (SCR001, SCR002, SCR003, SCR101, ...)
  BadComponentReferenceRule -> VIEW018
  ComponentReferenceValidationRule -> BIND005, SCR305
  ...
  NAME001, VIEW012, VIEW013, VIEW018 run on `**/view.json` files only, like ign-lint
  1 whitelisted file(s) would be linted, with their findings written to .iglint-baseline.json

What it does:

• Each ign-lint rule becomes its iglint rule with the same options and severity. Most ign-lint rules default to error where the iglint rule defaults to warning, so the migrated file sets severity = "error" for them, and the build fails on the same findings as before.
• Disabling PylintScriptRule disables the native SCR rules it maps to, so the migrated configuration reports what ign-lint did. Remove those enabled = false entries to get iglint's script checks, which is the main reason to switch.
• ign-lint lints only view.json files, so the rules it runs on top of iglint's defaults are selected for views only, in an [[overrides]] entry with files = ["**/view.json"]. Library modules, gateway events and WebDev scripts get iglint's native rules. Move a rule out of that entry (into extend-select) to run it everywhere.
• If an ign-lint whitelist exists (.ignition-lint-whitelist.txt or .whitelist.txt next to the configuration, or --whitelist FILE), the whitelisted files' current findings go into a baseline instead. Those files are linted from now on, and only their new findings fail.

Then write the files and gate on iglint check:

iglint migrate          # writes iglint.toml (and .iglint-baseline.json)
iglint baseline         # record today's findings of the other files too
iglint check

--output, --baseline and --force choose where the files go and whether existing files are overwritten.

Option 2: iglint ign-lint

iglint ign-lint [ARGS] accepts ign-lint 0.6.4's arguments (--config, --files, --whitelist, --ignore-warnings, --results-output, --fix, --generate-whitelist and the debug options) and prints ign-lint's output format, with ign-lint's node paths:

$ iglint ign-lint --config rule_config.json --files '**/view.json'
🔧 Loaded configuration from rule_config.json

📄 Evaluating file:
    projects/demo-hmi/com.inductiveautomation.perspective/views/Line/Overview/view.json

❌ Found 3 errors:

  PollingIntervalRule:
    • root.root.children[1].Clock.propConfig.props.text: 'dateFormat(now(1000), "HH:mm:ss")'


  ComponentReferenceValidationRule:
    • root.root.children[0].Title.propConfig.props.text: `view.custom.aera`: custom property `aera` is not defined


  UnusedCustomPropertiesRule:
    • custom.legacyFilter: custom property 'legacyFilter' is defined but never referenced

Called through a link named ign-lint, the binary behaves the same, so existing scripts keep working:

ln -s "$(command -v iglint)" /usr/local/bin/ign-lint
ign-lint --config rule_config.json --files "**/view.json"

There is also a drop-in pre-commit hook.

iglint check also reads a rule_config.json when no iglint.toml exists, with a deprecation warning.

Expect more findings

Caution

iglint ign-lint finds more than ign-lint, so a CI gate that passes with ign-lint can fail after the switch.

For example, BIND005 reports references to view custom properties and params that do not exist, which ign-lint's ComponentReferenceValidationRule does not check, and PylintScriptRule also runs iglint's own script rules. These extra findings are real: a reference that resolves nowhere evaluates to null with a bad quality.

iglint ign-lint has no baseline (ign-lint's --whitelist skips whole files). To adopt iglint without breaking the build, run iglint migrate, then iglint baseline, and gate on iglint check: today's findings are recorded and only new ones fail.

Intended differences

iglint reproduces ign-lint's findings except where ign-lint is wrong. Each difference is listed, with its reason, in Differences from ign-lint, and pinned by a parity test against output recorded from real ign-lint 0.6.4. The main ones:

• One invalid JSON file is reported (VIEW001) and the other files are still linted; ign-lint aborts.
• Deeply nested views are linted; ign-lint aborts the whole run with a RecursionError from about 250 nested components.
• References are matched exactly: {view.custom.foobar} does not count as a use of foo, and comments and strings never match a component reference pattern.
• Scripts are linted one by one, as Ignition compiles them, instead of as one combined file per view. Messages caused by the combined file (module-level naming, quote consistency across scripts) are gone, and names Ignition injects (system, the project's library packages) are not reported as undefined.
• PylintScriptRule findings depend on the installed pylint, so they are compared by message code, not as part of the parity test.

Speed

On the synthetic benchmark corpora, iglint ign-lint does the same work as ign-lint 7x (50 views) to 43x (20,000 views) faster with pylint, and 11x to 53x faster without, on a 16-CPU machine. On a production customer project of 287 views, with its own configuration and whitelist, iglint ign-lint was 14x faster than ign-lint's best configuration with pylint and 23x without. Details and the caveats are in Performance and caching.