kz930 opened a new issue, #7208:
URL: https://github.com/apache/texera/issues/7208
### What happened?
`DendrogramOpDesc.createDendrogram` interpolates the Color Threshold field
straight into scipy's `color_threshold` argument. The field is an
`EncodableString`, so what the template renders is a decode expression, and the
value reaching scipy at run time is a Python **string**. scipy compares it
against the linkage distances, which are float64, and numpy refuses. Every
non-empty Color Threshold therefore fails, whatever is typed; leaving the field
empty takes the `color_threshold=None` branch instead, which is why this has
gone unnoticed.
Expected: a numeric threshold colours the clusters joined below that linkage
distance.
Generated by `generatePythonCode()` for a threshold of `1.5` (`MS41` is its
base64):
fig = ff.create_dendrogram(data, labels=labels,
color_threshold=self.decode_python_template('MS41'))
Executing that generated module unmodified — only the `pytexera` import seam
stubbed, `decode_python_template` doing the real base64 decode — against a
five-row frame crashes. With the field left empty the same flow emits
`color_threshold=None` and produces the chart.
Straight against `ff.create_dendrogram` on plotly 5.24.1 and scipy 1.18.0,
the pinned versions:
| `color_threshold` | result |
| --- | --- |
| `None` — field empty | renders |
| `'1.5'` — string, what the operator emits today | **UFuncTypeError** |
| `1.5` — number | renders |
| `'default'` — scipy's own 0.7 × max distance | renders |
plotly passes the value through untouched, so a fix belongs on the Texera
side. It should convert with Python's `float()` and fall back to the original
string rather than casting unconditionally, so that `"default"` keeps working.
### How to reproduce?
1. Chain any source carrying two numeric columns and a label column into
Dendrogram.
2. Set Value X Column, Value Y Column and Labels.
3. Put `1.5` in Color Threshold and run — the operator fails with the error
below.
4. Clear Color Threshold and run the same workflow — it succeeds.
### Version/Branch
1.3.0-incubating-SNAPSHOT (main)
### Relevant log output
```shell
UFuncTypeError: ufunc 'greater_equal' did not contain a loop with signature
matching types
(<class 'numpy.dtypes.Float64DType'>, <class 'numpy.dtypes.StrDType'>) ->
None
```
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
