ponder-lab / Hybridize-Functions-Refactoring / #1647
72%

DEFAULT BRANCH: main
Ran
Jobs 1
Files 31
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: 72.219%. Remained the same
#1647

push

travis-pro

web-flow 
Extend precision audit to fixture 147 (no-tensor-parameter case) (#546)

* Extend precision audit to fixture 147 (no-tensor-parameter case)

Cover fixture 147 with an empty-tensor-types audit assertion. The Python uses `list = list()` followed by `list.append(tf.ones([1, 2]))` and `.append(tf.ones([2, 2]))`, then iterates: `for element in list: add(element, element)`. The dynamic-list construction (builtin `list()` plus `.append`, not a list literal) prevents Ariadne from tracking the appended tensors through to `element`, so neither parameter of `add` is classified as a tensor: the structural helper already expects `getHasTensorParameter() == false` here.

The audit-driver addition:

- **Layer 1:** assert `a.getTensorTypes()` and `b.getTensorTypes()` are both empty `Set`s.
- **Layer 2:** assert `inferInputSignature()` returns `Optional.empty()`—no tensor parameter means the inferred signature is dropped at the per-parameter classification step before reaching `inferSpec`.

This is the first audited fixture where the inferred signature legitimately doesn't exist (distinct from fixture 15's `Optional.empty()` which arose from dtype disagreement across call sites with classified-tensor parameters).

The Ariadne-side question (could a smarter analyzer track tensors through dynamic-list construction?) is a known precision gap area but out of scope for the per-fixture audit pinning—the runtime behavior of this fixture is documented as-emitted.

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

* Move fixture 147's commentary from inline to Javadoc

Following the convention from earlier audit batches: test-level documentation belongs in the method Javadoc, not as an inline comment in the body. Move the explanation of why neither parameter is classified as a tensor (dynamic-list construction via `list()` + `.append`) up into the Javadoc as a second paragraph after the issue reference.

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

---------... (continued)

1188 of 1645 relevant lines covered (72.22%)

0.72 hits per line

Jobs
ID Job ID Ran Files Coverage
1 #1647.1 31
72.22
Source Files on build #1647