ponder-lab / Hybridize-Functions-Refactoring / #1709 / 1
72%
main: 72%

DEFAULT BRANCH: main
Ran
Files 31
Run time 2s
Badge
Embed ▾
README BADGES
x

If you need to use a raster PNG badge, change the '.svg' to '.png' in the link

Markdown

Textile

RDoc

HTML

Rst
coverage: 72.276% (+0.02%) from 72.259%
#1709.1

push

travis-pro

web-flow 
Log a `WARNING` when an unknown `@tf.function` decorator argument is encountered (#558)

* Log a `WARNING` when an unknown `@tf.function` decorator argument is encountered

`Function.markParam` previously silently ignored decorator-argument
names it didn't recognize. The Javadoc on the switch justified the
choice as forward-compat with future TF versions, but silent-ignore
loses the signal—no test, evaluator output, or UI surface could see
that an unknown kwarg was used.

Add a `default` branch to the switch that logs the unknown argument
name (and the owning function) at `WARNING` level via the existing
`ILog LOG`. Forward-compat preserved (analysis still continues); the
signal is now surfaced in logs and CI output.

This is an intermediate step ahead of #204's original ask (custom
`UnknownHybridizationParameter` exception + matching test). The issue
stays open for the larger exception + test work.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

* Trim the unknown-argument log message

Two sentences with a rationale clause was verbose for a routine
warning. Cut to a single short line; the rationale lives in the
Javadoc on the switch.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

* Add period to the unknown-argument log message

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

* Add `testComputeParameters13` covering the unknown-kwarg WARN path

`@tf.function(unknown_argument=True)` exercises `markParam`'s `default` branch
(the WARN log added in this PR) without setting any recognized `*Param` flag.
The fixture is intentionally not Python-runnable—TF rejects unknown kwargs at
decoration time—so the three-check protocol's runtime step does not apply.
Analyzer-robustness on inputs TF itself rejects is precisely what the new
default branch is meant to guarantee.

* Switch `testComputeParameters13` fixture to a real TF kwarg

`experimental_attributes` is a real `tf.function` keyword argument added in TF
2.13 that Hybridize do... (continued)

1194 of 1652 relevant lines covered (72.28%)

0.72 hits per line

Source Files on job #1709.1