Jeet column-span mixin: transition from fixed 12‑column assumption to optional $context argument

Do not introduce a broad global compatibility flag. The recommended migration is to make $context explicit at call sites. If maintainers want a temporary bridge, expose a narrow, mixin-scoped legacy switch that only changes the fallback value used when $context is omitted.

Why the breakage happens

The breakage is most likely caused by default-context resolution. In the 6.x line, an omitted context behaved as if the parent were a fixed 12-column grid; in 7.x the same call may resolve against a different implicit count, so nested spans produce different percentage widths. What is confirmed is that percentage width is a function of both span and context. Any compatibility switch must therefore alter only the value used when $context is omitted; explicit contexts should remain identical.

Scoping the legacy switch

• Name the switch after the behavior, not a number: $jeet-column-span-legacy-context or $jeet-legacy-column-span.
• Default it to false and document it as a 7.x-only bridge.
• Restrict its effect to column-span when $context is omitted. Do not let it change column(), span(), or column-width unless those mixins are separately deprecated.
• Plan removal in the next major release so CSS output does not depend on hidden global state.

Deprecation warnings

Warnings should be compile-time and conditional:

1. Emit only when the legacy switch is true and $context is omitted.
2. Show the replacement call with an explicit context.
3. State the release in which the legacy switch will be removed.
4. Optionally emit once per stylesheet to avoid noise.

The one diagnostic that changes the advice

Confirm whether omitted $context was originally intended to mean the parent element's column count or a global default. If it meant parent inheritance, a flag is insufficient and call sites need explicit contexts.

Verification before naming anything

Treat the 6.x/7.x signatures as project context until verified against the installed package.

grep -rn "column-span" node_modules/jeet/

Build a minimal two-level fixture: a parent column-span(6) and a child column-span(1). Compile with the old and new Jeet versions and diff the emitted percentage widths. Repeat with the child passing an explicit $context; that output should not change. If a flag is implemented, compile with it off and on and assert that only omitted-context calls differ.

Assumptions

Sass and Stylus warning mechanisms and variable scoping differ. Implement the bridge only in the preprocessor flavor(s) Jeet actually ships, and test each one.