not working with VegaFusionWidget: Error: [object ArrayBuffer] is not serializable #46
Comments
|
This may be the same issue as #42 since VegaFusion also uses binary messages |
|
Bokeh 3.0 will have much better support for binary messages so we may just get this for free. |
|
@mattpap What do you think? Will your new serialization work allow serializing an ArrayBuffer embedded in some JSON object? |
|
It does have bidirectional support for |
|
I just stumbled on this repo today, because I was working on a reboot of flask_ipywidgets and was curious how you people were doing some similar work (the code looks very similar 👀 ). My guess is that https://github.com/bokeh/ipywidgets_bokeh/blob/main/ipywidgets_bokeh/kernel.py#L71 has an issue when there is a memoryview object with a non-byte format. A |
maartenbreddels
added a commit
to maartenbreddels/panel
that referenced
this issue
Feb 20, 2022
the offset should be in bytes, not elements. See also bokeh/ipywidgets_bokeh#46 (comment)
|
Thanks @maartenbreddels . I dont know the details by @philippjfr has said the next version of Bokeh will solve this issue. |
philippjfr
pushed a commit
to holoviz/panel
that referenced
this issue
Feb 20, 2022
the offset should be in bytes, not elements. See also bokeh/ipywidgets_bokeh#46 (comment)
philippjfr
pushed a commit
to holoviz/panel
that referenced
this issue
Mar 26, 2022
the offset should be in bytes, not elements. See also bokeh/ipywidgets_bokeh#46 (comment)
philippjfr
added a commit
to holoviz/panel
that referenced
this issue
Mar 27, 2022
* Fix jinja2 imports (#3258) * fix: wrong offset when memoryview format is non-byte (#3206) the offset should be in bytes, not elements. See also bokeh/ipywidgets_bokeh#46 (comment) * Support jupyter server root_dir with lab extension (#3172) Jupyter sends path relative to the root_dir which can be different from the cwd. This commit fixes the lab extension for panel preview to account for root_dir. Reference: #3170 * DOC: add note on enabling panel widget on Jupyter Lab (#3029) * Remove redundant and confusing JupyterLab install instructions (#3037) * Remove reference to Python 2 and link to Travis * Remove redondant install instructions to work with Jupyter * Update CHANGELOG * Add setuptools to build requirements * use https instead of the git:// protocol in GHA (#3092) * Try to switch to setuptools (#3138) * Try to switch to setuptools * Continue on error * CI: fix Macos tests workflow (#3160) * attempt to fix CI on macos * forgot bokeh channel * remove strict setting * reset strict and add pyctdev channel * remove continue_on_error and test py310 * clean up test workflow * add comment about PYCTDEV_SELF_CHANNEL * remove py310 * Revert change to build backend * Update setuptools pin * Pin setuptools * Further pin updates * Pin setuptools (#3265) * Fix image_url test * Pin pyvista * Bump panel.js version * Fix GH workflows * Bump panel.js version Co-authored-by: Maarten Breddels <[email protected]> Co-authored-by: Govinda Totla <[email protected]> Co-authored-by: Ray Bell <[email protected]> Co-authored-by: Maxime Liquet <[email protected]>
|
Wanted to update this issue with the current behavior using the latest versions of the various libraries: Versions: Example app: import panel as pn
import ipywidgets as ipw
pn.extension('ipywidgets')
from vega_datasets import data
import altair as alt
import vegafusion as vf
source = alt.UrlData(
data.flights_2k.url,
format={'parse': {'date': 'date'}}
)
brush = alt.selection_interval(encodings=['x'])
# Define the base chart, with the common parts of the
# background and highlights
base = alt.Chart(width=160, height=130).mark_bar().encode(
x=alt.X(alt.repeat('column')).bin(maxbins=20),
y='count()'
)
# gray background with selection
background = base.encode(
color=alt.value('#ddd')
).add_params(brush)
# blue highlights on the transformed data
highlight = base.transform_filter(brush)
# layer the two charts & repeat
chart = alt.layer(
background,
highlight,
data=source
).transform_calculate(
"time",
"hours(datum.date)"
).repeat(column=["distance", "delay", "time"])
chart_widget = vf.jupyter.VegaFusionWidget(chart)
pn.panel(chart_widget).servable()Run app: The window opens and the widgets loads successfully (as indicated by the VegaFusion logo being displayed). But an error occurs immediately after: Browser trace: Python trace It looks like the issue is related to the Here's the definitions of this prop on the Python side of the widget (See https://github.com/hex-inc/vegafusion/blob/main/python/vegafusion-jupyter/vegafusion_jupyter/widget.py#L31-L32) My guess is that the |
|
Thanks for the update @jonmmease! I'd love to support VegaFusion properly and we'll probably do two things to make that happen:
I'll reach out when we're ready to start on that. |
|
Just wanted to pass along an update. I've reworked the VegaFusion architecture a bit to make it possible to integrate VegaFusion into Altair's regular JupyterChart. See vega/altair#3281. This doesn't require binary serialization and already works well with Panel. I'm going to deprecate |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Panel: 0.12.6 vegafusion-jupyter: 0.0.2 vega-datasets 0.9.0 ipywidgets-bokeh 1.2.1
I'm trying to use the VegaFusionWidget ipywidget with Panel. See vega/vegafusion#62. But its raising the error below.
594.index.js:1 panicked at 'send_request function call failed: JsValue(Error: [object ArrayBuffer] is not serializable Error: [object ArrayBuffer] is not serializable at f.to_serializable (http://localhost:5006/static/js/bokeh.min.js?v=b3e0592a1e90448fa40dcddd5cd5217dcd6b3c7dd368020f6a8e4be4fdd1adbdeb2824eb75ab4f5120c80dc7684f8ff2aa2f8b00d6f393108c96d2f6f2cf0297:195:1391) at l.[serialize] (http://localhost:5006/static/js/bokeh.min.js?v=b3e0592a1e90448fa40dcddd5cd5217dcd6b3c7dd368020f6a8e4be4fdd1adbdeb2824eb75ab4f5120c80dc7684f8ff2aa2f8b00d6f393108c96d2f6f2cf0297:200:458) at f.to_serializable (http://localhost:5006/static/js/bokeh.min.js?v=b3e0592a1e90448fa40dcddd5cd5217dcd6b3c7dd368020f6a8e4be4fdd1adbdeb2824eb75ab4f5120c80dc7684f8ff2aa2f8b00d6f393108c96d2f6f2cf0297:195:1042) at f.to_serializable (http://localhost:5006/static/js/bokeh.min.js?v=b3e0592a1e90448fa40dcddd5cd5217dcd6b3c7dd368020f6a8e4be4fdd1adbdeb2824eb75ab4f5120c80dc7684f8ff2aa2f8b00d6f393108c96d2f6f2cf0297:195:1169) at E.create_json_patch (http://localhost:5006/static/js/bokeh.min.js?v=b3e0592a1e90448fa40dcddd5cd5217dcd6b3c7dd368020f6a8e4be4fdd1adbdeb2824eb75ab4f5120c80dc7684f8ff2aa2f8b00d6f393108c96d2f6f2cf0297:165:9308) at r._document_changed (http://localhost:5006/static/js/bokeh.min.js?v=b3e0592a1e90448fa40dcddd5cd5217dcd6b3c7dd368020f6a8e4be4fdd1adbdeb2824eb75ab4f5120c80dc7684f8ff2aa2f8b00d6f393108c96d2f6f2cf0297:590:850) at _document_listener (http://localhost:5006/static/js/bokeh.min.js?v=b3e0592a1e90448fa40dcddd5cd5217dcd6b3c7dd368020f6a8e4be4fdd1adbdeb2824eb75ab4f5120c80dc7684f8ff2aa2f8b00d6f393108c96d2f6f2cf0297:590:162) at E._trigger_on_change (http://localhost:5006/static/js/bokeh.min.js?v=b3e0592a1e90448fa40dcddd5cd5217dcd6b3c7dd368020f6a8e4be4fdd1adbdeb2824eb75ab4f5120c80dc7684f8ff2aa2f8b00d6f393108c96d2f6f2cf0297:165:4698) at gg.bk_send (http://localhost:5006/static/extensions/ipywidgets_bokeh/ipywidgets_bokeh.js?v=206620a657f1ccc8af72b866fdcef20e641c48f39ce27f338ce8313949c63e53:2:2745297) at e.send (http://localhost:5006/static/extensions/ipywidgets_bokeh/ipywidgets_bokeh.js?v=206620a657f1ccc8af72b866fdcef20e641c48f39ce27f338ce8313949c63e53:2:2743285))', vegafusion-wasm/src/lib.rs:330:14 Stack: Error at Object.G (https://unpkg.com/[email protected]/dist/594.index.js:1:7607) at __wbg_new_693216e109162396 (https://unpkg.com/vegafusion-jupyter@%5E0.0.2/dist/index.js:1:22417) at https://unpkg.com/[email protected]/dist/d0048f4ff7544ce1ea76.module.wasm:wasm-function[2561]:0x3d8a32 at https://unpkg.com/[email protected]/dist/d0048f4ff7544ce1ea76.module.wasm:wasm-function[5005]:0x4491e5 at https://unpkg.com/[email protected]/dist/d0048f4ff7544ce1ea76.module.wasm:wasm-function[3646]:0x42ed5b at https://unpkg.com/[email protected]/dist/d0048f4ff7544ce1ea76.module.wasm:wasm-function[3935]:0x43bbbf at https://unpkg.com/[email protected]/dist/d0048f4ff7544ce1ea76.module.wasm:wasm-function[4371]:0x445b30 at https://unpkg.com/[email protected]/dist/d0048f4ff7544ce1ea76.module.wasm:wasm-function[4417]:0x4463d7 at https://unpkg.com/[email protected]/dist/d0048f4ff7544ce1ea76.module.wasm:wasm-function[3974]:0x43ceec at https://unpkg.com/[email protected]/dist/d0048f4ff7544ce1ea76.module.wasm:wasm-function[1422]:0x33ce35 nn @ 594.index.js:1 ipywidgets_bokeh.js?v=206620a657f1ccc8af72b866fdcef20e641c48f39ce27f338ce8313949c63e53:2 Error: Could not create a view for model id 0582b62fd2c64638b4821fd5d8c4040d at ipywidgets_bokeh.js?v=206620a657f1ccc8af72b866fdcef20e641c48f39ce27f338ce8313949c63e53:2:2087124 (anonymous) @ ipywidgets_bokeh.js?v=206620a657f1ccc8af72b866fdcef20e641c48f39ce27f338ce8313949c63e53:2 ipywidgets_bokeh.js?v=206620a657f1ccc8af72b866fdcef20e641c48f39ce27f338ce8313949c63e53:2 Error: Could not create view at ipywidgets_bokeh.js?v=206620a657f1ccc8af72b866fdcef20e641c48f39ce27f338ce8313949c63e53:2:2087124 at async gg.render (ipywidgets_bokeh.js?v=206620a657f1ccc8af72b866fdcef20e641c48f39ce27f338ce8313949c63e53:2:2744026) at async _g._render (ipywidgets_bokeh.js?v=206620a657f1ccc8af72b866fdcef20e641c48f39ce27f338ce8313949c63e53:2:2744950) (anonymous) @ ipywidgets_bokeh.js?v=206620a657f1ccc8af72b866fdcef20e641c48f39ce27f338ce8313949c63e53:2 2d0048f4ff7544ce1ea76.module.wasm:0x42ed91 Uncaught (in promise) RuntimeError: unreachable at d0048f4ff7544ce1ea76.module.wasm:0x42ed91 at d0048f4ff7544ce1ea76.module.wasm:0x43bbbf at d0048f4ff7544ce1ea76.module.wasm:0x445b30 at d0048f4ff7544ce1ea76.module.wasm:0x4463d7 at d0048f4ff7544ce1ea76.module.wasm:0x43ceec at d0048f4ff7544ce1ea76.module.wasm:0x33ce35 at d0048f4ff7544ce1ea76.module.wasm:0x1714ab at d0048f4ff7544ce1ea76.module.wasm:0x43abec at d.J [as render_vegafusion] (594.index.js:1:4972) at d.value_changed (index.js:1:14547)The text was updated successfully, but these errors were encountered: