ponder-lab / Hybridize-Functions-Refactoring / #2252
79%
main: 80%

LAST BUILD BRANCH: gh-readonly-queue/main/pr-739-6721f597b1febe25181379bb3e6a99b1f85dd4e0
DEFAULT BRANCH: main
Ran
Jobs 1
Files 32
Run time 1min
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: 79.286%. Remained the same
#2252

push

github

web-flow 
Pin the subscript-slice tensor over-typing (wala/ML#656) (#708)

* Pin the np.array and subscript-slice tensor over-typings (wala/ML#651, #656)

Two regression guards that confirm the over-typing mechanisms and flip when
fixed:
- `testTfTrainNumpyOverTyping`: an `np.array` fed only to `tf.train.*` protobuf
  builders types the parameter as a tensor (wala/ML#651).
- `testSubscriptSliceOpaqueOverTyping`: a subscript-slice of an opaque argparse
  attribute types the parameter as a tensor; a control (the same attribute
  unsliced) is correctly not typed, isolating the slice as the trigger
  (wala/ML#656).

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01FuhM8SsTRB6BraLt3Ph7ZC

* Drop the np.array pin; wala/ML#651 is not an over-typing bug

wala/ML#651 was reclassified: an `np.ndarray` is a tensor-like value, so Ariadne
correctly types the parameter as a tensor. `serialize`'s real defect is that its
body performs no tensor computation, a Hybridize advantage-analysis gap tracked
in #709, not an Ariadne over-typing. The `testTfTrainNumpyOverTyping` pin
asserted correct behavior framed as a bug, so it is removed; the
serialize/no-tensor-op case moves to the #709 work.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01FuhM8SsTRB6BraLt3Ph7ZC

---------

Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

1665 of 2100 relevant lines covered (79.29%)

0.79 hits per line

Jobs
ID Job ID Ran Files Coverage
1 #2252.1 32
79.29
Source Files on build #2252