| |
|
|
|
|
|
|
| |
Re-enabling native-comp surfaced a suite-wide fragility. When a test redefines a C primitive (or a native-compiled function), native-comp routes native callers through a trampoline that calls the mock with the primitive's maximum arity. A fixed-arity mock narrower than the primitive then throws wrong-number-of-arguments, intermittently, as the eln-cache fills.
I swept every arity-narrow subr mock to append &rest _ (188 sites, preserving any named args the body uses), and added tests/test-meta-subr-mock-arity.el, which fails make test on any subr mock too narrow for the primitive's arity. The rule isn't "never mock a subr". The suite mocks message and completing-read freely. It's "a subr mock must accept the primitive's arity."
Background, the three failure modes, and the research are in docs/native-comp-subr-mocking.org.
|
|
|
Two test files for keybindings.el. cj/jump-open-var gets full N/B/E
coverage (6 tests): existing-file happy path, plus error paths for
unbound symbol, nil value, non-string value, empty string, and missing
file. The smoke file for the auto-generated cj/jump-to-NAME commands
asserts that each spec entry has an fbound command, that the command
is bound in cj/jump-map at the spec's key, that calling each command
invokes cj/jump-open-var with the spec's var, and that cj/jump-map is
mounted under cj/custom-keymap at "j".
The test fixture variable is declared at top level. If it were
let-bound inside a test under lexical-binding, the let would create a
lexical binding that shadows the dynamic one. The production code's
symbol-value would then miss what setq writes. find-file is mocked at
the boundary so the existing-file test doesn't actually open a buffer.
10 tests pass. No production change in keybindings.el.
|
