ponder-lab / Hybridize-Functions-Refactoring / #3661
90%

DEFAULT BRANCH: main
Ran
Jobs 1
Files 48
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: 88.632% (-0.02%) from 88.649%
#3661

push

github

web-flow 
Bump Ariadne to 0.52.107 (#1106)

* Build against Ariadne's 0.52.107-SNAPSHOT

The snapshot is the one Ariadne's continuous integration publishes from its master branch, so this branch's suite runs against the engine the next release will ship. Tycho 5.0.4 wraps a remote timestamped snapshot under a version below the one this project imports, so the target restores the base-version wrap instructions until the release replaces the snapshot.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01NTnpjPjJj2aZy1ZkUD1eAn

* Advance five pins to Ariadne 0.52.107's behavior

- `maps_caller` (testInvalidNameArgumentCallers), `wild_map_fn` (testUnresolvedStaticDimensionThroughClosure) and `via_map_fn` (testNumpyCallsThroughClosure): the function `tf.map_fn` is given is now called with an element of `elems` (wala/ML#942), so each callback has a call-graph node and each function joins its test's refused or withheld cases. Each raises under `tf.function` on TensorFlow 2.9.3.
- `BatchNormalization.call` (testKerasBuiltinLayerSubclassSuppliedParameters) and `Affine.call` (testKerasBuiltinDenseSubclassSuppliedParameters): the built-in layer summaries now declare `Layer` as their superclass (wala/ML#1098), so `training` is recognized as supplied by Keras and the signature is withheld with DEFAULTED_PARAMETER_SUPPLIED. The signature inferred before raises on the first call on TensorFlow 2.9.3.

Each TODO these pins carried is removed, since each pin now asserts the correct behavior.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01NTnpjPjJj2aZy1ZkUD1eAn

* Pin a layer's `add_loss` as a Python side-effect

Ariadne now models `Layer.add_loss` and `Model.add_loss` as writes to the layer's losses (https://github.com/wala/ML/issues/1095), so a Keras `call` that registers a loss writes an object that outlives the call. The new test pins that `Var.call`, `Outer.call`, ... (continued)

5294 of 5973 relevant lines covered (88.63%)

0.89 hits per line

Coverage Regressions

Lines Coverage ∆ File
1
94.41
-0.21% edu.cuny.hunter.hybridize.core/src/edu/cuny/hunter/hybridize/core/analysis/StaticShapeReadAnalysis.java
Jobs
ID Job ID Ran Files Coverage
1 #3661.1 48
88.63
Source Files on build #3661