Bokeh widgets without bokeh server

Andy_Baker · March 28, 2016, 10:46am

Hi,

I’m using bokeh.embed.components to generate bokeh plots inside a Django app. I’m really keen to avoid using bokeh server due to the increase in deployment and maintenance complexity.

I understand that Bokeh has limited functionality without the server but the docs aren’t terribly clear on exactly what is and isn’t possible. I wanted to generate some widgets and then figure out how to control a client-side plot using them. It seems tantalizingly close to being possible.

Am I barking up the wrong tree? Is serverless bokeh something that you consider a viable setup or is it more of a curiosity?

Luke_Canavan1 · March 28, 2016, 4:29pm

It’s entirely possible to use Bokeh widgets in a traditional web framework. Generating the widgets is pretty straightforward, the only additional work is to add the interactivity that you want the widgets to perform. There are several Bokeh-native ways to add callbacks (http://bokeh.pydata.org/en/latest/docs/user_guide/interaction.html#javascript-callbacks) or you can write your own client-side javascript and on-change listener.

···

On Monday, March 28, 2016 at 6:53:05 AM UTC-5, Andy Baker wrote:

Hi,

I’m using bokeh.embed.components to generate bokeh plots inside a Django app. I’m really keen to avoid using bokeh server due to the increase in deployment and maintenance complexity.

I understand that Bokeh has limited functionality without the server but the docs aren’t terribly clear on exactly what is and isn’t possible. I wanted to generate some widgets and then figure out how to control a client-side plot using them. It seems tantalizingly close to being possible.

Am I barking up the wrong tree? Is serverless bokeh something that you consider a viable setup or is it more of a curiosity?

Andy_Baker · March 30, 2016, 8:09am

In that case I’m doing something stupid. Something like this:

def widget_test(request):

    charts = [Button(label="hello")]

    scripts, charts = components(charts, CDN)

    return TemplateResponse(
        request,
        'bar_chart.html', {"scripts": scripts, "charts": charts,},
    )

(which works perfectly when I swap in a Chart for a button) gives:

Module `Button’ does not exists. The problem may be two fold. Either a model was requested that’s available in an extra bundle, e.g. a widget, or a custom model was requested, but it wasn’t registered before first usage.’

I’ve tried loading bokeh_widgets.js - this gives a different error.

And I’ve tried adding the button to various layout elements instead of passing it directly to components() and I get a similar error.

Is there any example code I could see?

···

On Monday, 28 March 2016 17:29:37 UTC+1, [email protected] wrote:

It’s entirely possible to use Bokeh widgets in a traditional web framework. Generating the widgets is pretty straightforward, the only additional work is to add the interactivity that you want the widgets to perform. There are several Bokeh-native ways to add callbacks (http://bokeh.pydata.org/en/latest/docs/user_guide/interaction.html#javascript-callbacks) or you can write your own client-side javascript and on-change listener.

On Monday, March 28, 2016 at 6:53:05 AM UTC-5, Andy Baker wrote:

Hi,

I’m using bokeh.embed.components to generate bokeh plots inside a Django app. I’m really keen to avoid using bokeh server due to the increase in deployment and maintenance complexity.

I understand that Bokeh has limited functionality without the server but the docs aren’t terribly clear on exactly what is and isn’t possible. I wanted to generate some widgets and then figure out how to control a client-side plot using them. It seems tantalizingly close to being possible.

Am I barking up the wrong tree? Is serverless bokeh something that you consider a viable setup or is it more of a curiosity?

Luke_Canavan1 · March 30, 2016, 2:16pm

This NYU project uses Bokeh plots served as part of a cherrypy web application:

Here’s the app route that’s serving the plot:

https://github.com/ViDA-NYU/domain_discovery_tool/blob/canavandl/domain_xfilter/vis/server.py#L375

Here’s the template that it’s rendering into:

https://github.com/ViDA-NYU/domain_discovery_tool/blob/canavandl/domain_xfilter/vis/html/cross_filter.html

Looking at your code. I wouldn’t include the CDN in your components method call (I think that’s a deprecated API and doesn’t do anything). Then you’ll have to make sure that you’re loading bokeh-widgets resources (JS and CSS) as a style and script in your template, either from a local directory or CDN.

Resources docs:

http://bokeh.pydata.org/en/latest/docs/user_guide/embed.html#components

As for the error message you’re seeing, it happens when you have a bokeh component that doesn’t have a corresponding BokehJS component. It sometimes happens if you have a mismatch between Bokeh and BokehJS versions, so that might be something to check.

Let us know if that doesn’t work.

lukec

···

On Wednesday, March 30, 2016 at 3:09:42 AM UTC-5, Andy Baker wrote:

In that case I’m doing something stupid. Something like this:

def widget_test(request):

    charts = [Button(label="hello")]

    scripts, charts = components(charts, CDN)

    return TemplateResponse(
        request,
        'bar_chart.html', {"scripts": scripts, "charts": charts,},
    )

(which works perfectly when I swap in a Chart for a button) gives:

Module `Button’ does not exists. The problem may be two fold. Either a model was requested that’s available in an extra bundle, e.g. a widget, or a custom model was requested, but it wasn’t registered before first usage.’

I’ve tried loading bokeh_widgets.js - this gives a different error.

And I’ve tried adding the button to various layout elements instead of passing it directly to components() and I get a similar error.

Is there any example code I could see?

On Monday, 28 March 2016 17:29:37 UTC+1, [email protected] wrote:

It’s entirely possible to use Bokeh widgets in a traditional web framework. Generating the widgets is pretty straightforward, the only additional work is to add the interactivity that you want the widgets to perform. There are several Bokeh-native ways to add callbacks (http://bokeh.pydata.org/en/latest/docs/user_guide/interaction.html#javascript-callbacks) or you can write your own client-side javascript and on-change listener.

On Monday, March 28, 2016 at 6:53:05 AM UTC-5, Andy Baker wrote:

Hi,

I’m using bokeh.embed.components to generate bokeh plots inside a Django app. I’m really keen to avoid using bokeh server due to the increase in deployment and maintenance complexity.

I understand that Bokeh has limited functionality without the server but the docs aren’t terribly clear on exactly what is and isn’t possible. I wanted to generate some widgets and then figure out how to control a client-side plot using them. It seems tantalizingly close to being possible.

Am I barking up the wrong tree? Is serverless bokeh something that you consider a viable setup or is it more of a curiosity?

Andy_Baker · March 31, 2016, 1:14pm

Marvellous. Even the callbacks work.

I was so close but the docs led me astray. All I needed was:

def bar_chart(request):

bokeh_script, charts = components(Button(label='hello'))

return TemplateResponse(
    request,
    'bar_chart.html',
    {'bokeh_script': bokeh_script, 'charts': charts,}
)

and this in my template:

```

···

On Wednesday, 30 March 2016 15:16:40 UTC+1, [email protected] wrote:

This NYU project uses Bokeh plots served as part of a cherrypy web application:

Here’s the app route that’s serving the plot:

https://github.com/ViDA-NYU/domain_discovery_tool/blob/canavandl/domain_xfilter/vis/server.py#L375

Here’s the template that it’s rendering into:

https://github.com/ViDA-NYU/domain_discovery_tool/blob/canavandl/domain_xfilter/vis/html/cross_filter.html

Looking at your code. I wouldn’t include the CDN in your components method call (I think that’s a deprecated API and doesn’t do anything). Then you’ll have to make sure that you’re loading bokeh-widgets resources (JS and CSS) as a style and script in your template, either from a local directory or CDN.

Resources docs:

http://bokeh.pydata.org/en/latest/docs/user_guide/embed.html#components

As for the error message you’re seeing, it happens when you have a bokeh component that doesn’t have a corresponding BokehJS component. It sometimes happens if you have a mismatch between Bokeh and BokehJS versions, so that might be something to check.

Let us know if that doesn’t work.

lukec

On Wednesday, March 30, 2016 at 3:09:42 AM UTC-5, Andy Baker wrote:

In that case I’m doing something stupid. Something like this:

def widget_test(request):

    charts = [Button(label="hello")]

    scripts, charts = components(charts, CDN)

    return TemplateResponse(
        request,
        'bar_chart.html', {"scripts": scripts, "charts": charts,},
    )

(which works perfectly when I swap in a Chart for a button) gives:

Module `Button’ does not exists. The problem may be two fold. Either a model was requested that’s available in an extra bundle, e.g. a widget, or a custom model was requested, but it wasn’t registered before first usage.’

I’ve tried loading bokeh_widgets.js - this gives a different error.

And I’ve tried adding the button to various layout elements instead of passing it directly to components() and I get a similar error.

Is there any example code I could see?

On Monday, 28 March 2016 17:29:37 UTC+1, [email protected] wrote:

It’s entirely possible to use Bokeh widgets in a traditional web framework. Generating the widgets is pretty straightforward, the only additional work is to add the interactivity that you want the widgets to perform. There are several Bokeh-native ways to add callbacks (http://bokeh.pydata.org/en/latest/docs/user_guide/interaction.html#javascript-callbacks) or you can write your own client-side javascript and on-change listener.

On Monday, March 28, 2016 at 6:53:05 AM UTC-5, Andy Baker wrote:

Hi,

I’m using bokeh.embed.components to generate bokeh plots inside a Django app. I’m really keen to avoid using bokeh server due to the increase in deployment and maintenance complexity.

I understand that Bokeh has limited functionality without the server but the docs aren’t terribly clear on exactly what is and isn’t possible. I wanted to generate some widgets and then figure out how to control a client-side plot using them. It seems tantalizingly close to being possible.

Am I barking up the wrong tree? Is serverless bokeh something that you consider a viable setup or is it more of a curiosity?

Bryan · April 2, 2016, 9:27pm

Hi Andy,

can you elaborate on:

I was so close but the docs led me astray.

This is an especially important features, so if there is something misleading or missing about the docs, we would want to improve them.

Thanks,

Bryan

···

On Mar 31, 2016, at 8:14 AM, Andy Baker <[email protected]> wrote:

Marvellous. Even the callbacks work.

I was so close but the docs led me astray. All I needed was:

def bar_chart(request):

    bokeh_script, charts = components(Button(label='hello'))

    return TemplateResponse(
        request,
        'bar_chart.html',
        {'bokeh_script': bokeh_script, 'charts': charts,}
    )

and this in my template:

<link href="http://cdn.pydata.org/bokeh/release/bokeh-0.11.0.css" rel="stylesheet">
<link href="http://cdn.pydata.org/bokeh/release/bokeh-widgets-0.11.0.css" rel="stylesheet">
<script src="http://cdn.pydata.org/bokeh/release/bokeh-0.11.0.js"></script>
<script src="http://cdn.pydata.org/bokeh/release/bokeh-widgets-0.11.0.js"></script>

On Wednesday, 30 March 2016 15:16:40 UTC+1, lcan...@continuum.io wrote:
This NYU project uses Bokeh plots served as part of a cherrypy web application:

Here's the app route that's serving the plot:
https://github.com/ViDA-NYU/domain_discovery_tool/blob/canavandl/domain_xfilter/vis/server.py#L375

Here's the template that it's rendering into:
https://github.com/ViDA-NYU/domain_discovery_tool/blob/canavandl/domain_xfilter/vis/html/cross_filter.html

Looking at your code. I wouldn't include the CDN in your `components` method call (I think that's a deprecated API and doesn't do anything). Then you'll have to make sure that you're loading bokeh-widgets resources (JS and CSS) as a style and script in your template, either from a local directory or CDN.

Resources docs:
http://bokeh.pydata.org/en/latest/docs/user_guide/embed.html#components

As for the error message you're seeing, it happens when you have a bokeh component that doesn't have a corresponding BokehJS component. It sometimes happens if you have a mismatch between Bokeh and BokehJS versions, so that might be something to check.

Let us know if that doesn't work.

lukec

On Wednesday, March 30, 2016 at 3:09:42 AM UTC-5, Andy Baker wrote:
In that case I'm doing something stupid. Something like this:

def widget_test(request):

    charts = [Button(label="hello")]

    scripts, charts = components(charts, CDN)

    return TemplateResponse(
        request,
        'bar_chart.html', {"scripts": scripts, "charts": charts,},
    )

(which works perfectly when I swap in a Chart for a button) gives:

   Module `Button' does not exists. The problem may be two fold. Either a model was requested that's available in an extra bundle, e.g. a widget, or a custom model was requested, but it wasn't registered before first usage.'

I've tried loading bokeh_widgets.js - this gives a different error.

And I've tried adding the button to various layout elements instead of passing it directly to components() and I get a similar error.

Is there any example code I could see?

On Monday, 28 March 2016 17:29:37 UTC+1, lcan...@continuum.io wrote:
It's entirely possible to use Bokeh widgets in a traditional web framework. Generating the widgets is pretty straightforward, the only additional work is to add the interactivity that you want the widgets to perform. There are several Bokeh-native ways to add callbacks (Interaction — Bokeh 3.3.2 Documentation) or you can write your own client-side javascript and on-change listener.

On Monday, March 28, 2016 at 6:53:05 AM UTC-5, Andy Baker wrote:
Hi,

I'm using bokeh.embed.components to generate bokeh plots inside a Django app. I'm really keen to avoid using bokeh server due to the increase in deployment and maintenance complexity.

I understand that Bokeh has limited functionality without the server but the docs aren't terribly clear on exactly what is and isn't possible. I wanted to generate some widgets and then figure out how to control a client-side plot using them. It seems tantalizingly close to being possible.

Am I barking up the wrong tree? Is serverless bokeh something that you consider a viable setup or is it more of a curiosity?

--
You received this message because you are subscribed to the Google Groups "Bokeh Discussion - Public" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [email protected].
To post to this group, send email to [email protected].
To view this discussion on the web visit https://groups.google.com/a/continuum.io/d/msgid/bokeh/4284d9ba-5a4b-469a-9bd5-3c4eba2eaf02%40continuum.io\.
For more options, visit https://groups.google.com/a/continuum.io/d/optout\.

Barnabas_Szabolcs · April 16, 2016, 2:53pm

Hi Bryan,

I just checked, I’m using bokeh 0.11.1 py27_0, part of anaconda and the script is supplied from http://cdn.pydata.org/bokeh/release/bokeh-0.11.1.min.js. It still says
Uncaught Error: Module `Slider’ does not exists. The problem may be two fold. Either a model was requested that’s available in an extra bundle, e.g. a widget, or a custom model was requested, but it wasn’t registered before first usage.

Do I need to include some extra js script other than this and what bokeh gives from components?

Thanks, B.

···

On Saturday, 2 April 2016 23:27:17 UTC+2, Bryan Van de ven wrote:

Hi Andy,

can you elaborate on:

I was so close but the docs led me astray.

This is an especially important features, so if there is something misleading or missing about the docs, we would want to improve them.

Thanks,

Bryan

On Mar 31, 2016, at 8:14 AM, Andy Baker [email protected] wrote:

Marvellous. Even the callbacks work.

I was so close but the docs led me astray. All I needed was:

def bar_chart(request):
bokeh_script, charts = components(Button(label='hello'))
return TemplateResponse(
    request,
    'bar_chart.html',
    {'bokeh_script': bokeh_script, 'charts': charts,}
)
and this in my template:

On Wednesday, 30 March 2016 15:16:40 UTC+1, [email protected] wrote:

This NYU project uses Bokeh plots served as part of a cherrypy web application:

Here’s the app route that’s serving the plot:

https://github.com/ViDA-NYU/domain_discovery_tool/blob/canavandl/domain_xfilter/vis/server.py#L375

Here’s the template that it’s rendering into:

https://github.com/ViDA-NYU/domain_discovery_tool/blob/canavandl/domain_xfilter/vis/html/cross_filter.html

Looking at your code. I wouldn’t include the CDN in your components method call (I think that’s a deprecated API and doesn’t do anything). Then you’ll have to make sure that you’re loading bokeh-widgets resources (JS and CSS) as a style and script in your template, either from a local directory or CDN.

Resources docs:

http://bokeh.pydata.org/en/latest/docs/user_guide/embed.html#components

As for the error message you’re seeing, it happens when you have a bokeh component that doesn’t have a corresponding BokehJS component. It sometimes happens if you have a mismatch between Bokeh and BokehJS versions, so that might be something to check.

Let us know if that doesn’t work.

lukec

On Wednesday, March 30, 2016 at 3:09:42 AM UTC-5, Andy Baker wrote:

In that case I’m doing something stupid. Something like this:

def widget_test(request):
charts = [Button(label="hello")]
scripts, charts = components(charts, CDN)
return TemplateResponse(
    request,
    'bar_chart.html', {"scripts": scripts, "charts": charts,},
)
(which works perfectly when I swap in a Chart for a button) gives:

Module `Button’ does not exists. The problem may be two fold. Either a model was requested that’s available in an extra bundle, e.g. a widget, or a custom model was requested, but it wasn’t registered before first usage.’

I’ve tried loading bokeh_widgets.js - this gives a different error.

And I’ve tried adding the button to various layout elements instead of passing it directly to components() and I get a similar error.

Is there any example code I could see?

On Monday, 28 March 2016 17:29:37 UTC+1, [email protected] wrote:

It’s entirely possible to use Bokeh widgets in a traditional web framework. Generating the widgets is pretty straightforward, the only additional work is to add the interactivity that you want the widgets to perform. There are several Bokeh-native ways to add callbacks (http://bokeh.pydata.org/en/latest/docs/user_guide/interaction.html#javascript-callbacks) or you can write your own client-side javascript and on-change listener.

On Monday, March 28, 2016 at 6:53:05 AM UTC-5, Andy Baker wrote:

Hi,

I’m using bokeh.embed.components to generate bokeh plots inside a Django app. I’m really keen to avoid using bokeh server due to the increase in deployment and maintenance complexity.

I understand that Bokeh has limited functionality without the server but the docs aren’t terribly clear on exactly what is and isn’t possible. I wanted to generate some widgets and then figure out how to control a client-side plot using them. It seems tantalizingly close to being possible.

Am I barking up the wrong tree? Is serverless bokeh something that you consider a viable setup or is it more of a curiosity?

–
You received this message because you are subscribed to the Google Groups “Bokeh Discussion - Public” group.

To unsubscribe from this group and stop receiving emails from it, send an email to [email protected].

To post to this group, send email to [email protected].

To view this discussion on the web visit https://groups.google.com/a/continuum.io/d/msgid/bokeh/4284d9ba-5a4b-469a-9bd5-3c4eba2eaf02%40continuum.io.

For more options, visit https://groups.google.com/a/continuum.io/d/optout.

Bryan · April 16, 2016, 5:46pm

Hi Barnabas,

In order to reduce the page load when widgets are not used, BokehJS was recently split up into a few pieces that could be used only as-needed. If you have widgets in your document, there also needs to be a script tag for

http://cdn.pydata.org/bokeh/release/bokeh-widgets-0.11.1.min.js

If you use things like output_file, or any of the "autoload" functions, this is typically handled for you. But bokeh.embed.components is a bit lower-level, and assumes the proper resources are already present. So you need to add the script tag to your page template, most likely.

I note that the docs here:

bokeh.embed — Bokeh 3.3.2 Documentation

Could probably stand to be more explicit about the different resources that need to be loaded. Can you make an issue on the GitHub issue tracker to improve this docstring with more information?

Thanks,

Bryan

···

On Apr 16, 2016, at 9:53 AM, [email protected] wrote:

Hi Bryan,

I just checked, I'm using bokeh 0.11.1 py27_0, part of anaconda and the script is supplied from http://cdn.pydata.org/bokeh/release/bokeh-0.11.1.min.js\. It still says
Uncaught Error: Module `Slider' does not exists. The problem may be two fold. Either a model was requested that's available in an extra bundle, e.g. a widget, or a custom model was requested, but it wasn't registered before first usage.

Do I need to include some extra js script other than this and what bokeh gives from components?

Thanks, B.

On Saturday, 2 April 2016 23:27:17 UTC+2, Bryan Van de ven wrote:
Hi Andy,

can you elaborate on:

> I was so close but the docs led me astray.

This is an especially important features, so if there is something misleading or missing about the docs, we would want to improve them.

Thanks,

Bryan

> On Mar 31, 2016, at 8:14 AM, Andy Baker <[email protected]> wrote:
>
> Marvellous. Even the callbacks work.
>
> I was so close but the docs led me astray. All I needed was:
>
> def bar_chart(request):
>
> bokeh_script, charts = components(Button(label='hello'))
>
> return TemplateResponse(
> request,
> 'bar_chart.html',
> {'bokeh_script': bokeh_script, 'charts': charts,}
> )
>
> and this in my template:
>
> <link href="http://cdn.pydata.org/bokeh/release/bokeh-0.11.0.css" rel="stylesheet">
> <link href="http://cdn.pydata.org/bokeh/release/bokeh-widgets-0.11.0.css" rel="stylesheet">
> <script src="http://cdn.pydata.org/bokeh/release/bokeh-0.11.0.js"></script>
> <script src="http://cdn.pydata.org/bokeh/release/bokeh-widgets-0.11.0.js"></script>
>
> On Wednesday, 30 March 2016 15:16:40 UTC+1, lcan...@continuum.io wrote:
> This NYU project uses Bokeh plots served as part of a cherrypy web application:
>
> Here's the app route that's serving the plot:
> https://github.com/ViDA-NYU/domain_discovery_tool/blob/canavandl/domain_xfilter/vis/server.py#L375
>
> Here's the template that it's rendering into:
> https://github.com/ViDA-NYU/domain_discovery_tool/blob/canavandl/domain_xfilter/vis/html/cross_filter.html
>
> Looking at your code. I wouldn't include the CDN in your `components` method call (I think that's a deprecated API and doesn't do anything). Then you'll have to make sure that you're loading bokeh-widgets resources (JS and CSS) as a style and script in your template, either from a local directory or CDN.
>
> Resources docs:
> http://bokeh.pydata.org/en/latest/docs/user_guide/embed.html#components
>
> As for the error message you're seeing, it happens when you have a bokeh component that doesn't have a corresponding BokehJS component. It sometimes happens if you have a mismatch between Bokeh and BokehJS versions, so that might be something to check.
>
> Let us know if that doesn't work.
>
> lukec
>
> On Wednesday, March 30, 2016 at 3:09:42 AM UTC-5, Andy Baker wrote:
> In that case I'm doing something stupid. Something like this:
>
> def widget_test(request):
>
> charts = [Button(label="hello")]
>
> scripts, charts = components(charts, CDN)
>
> return TemplateResponse(
> request,
> 'bar_chart.html', {"scripts": scripts, "charts": charts,},
> )
>
> (which works perfectly when I swap in a Chart for a button) gives:
>
> Module `Button' does not exists. The problem may be two fold. Either a model was requested that's available in an extra bundle, e.g. a widget, or a custom model was requested, but it wasn't registered before first usage.'
>
> I've tried loading bokeh_widgets.js - this gives a different error.
>
> And I've tried adding the button to various layout elements instead of passing it directly to components() and I get a similar error.
>
> Is there any example code I could see?
>
> On Monday, 28 March 2016 17:29:37 UTC+1, lcan...@continuum.io wrote:
> It's entirely possible to use Bokeh widgets in a traditional web framework. Generating the widgets is pretty straightforward, the only additional work is to add the interactivity that you want the widgets to perform. There are several Bokeh-native ways to add callbacks (Interaction — Bokeh 3.3.2 Documentation) or you can write your own client-side javascript and on-change listener.
>
> On Monday, March 28, 2016 at 6:53:05 AM UTC-5, Andy Baker wrote:
> Hi,
>
> I'm using bokeh.embed.components to generate bokeh plots inside a Django app. I'm really keen to avoid using bokeh server due to the increase in deployment and maintenance complexity.
>
> I understand that Bokeh has limited functionality without the server but the docs aren't terribly clear on exactly what is and isn't possible. I wanted to generate some widgets and then figure out how to control a client-side plot using them. It seems tantalizingly close to being possible.
>
> Am I barking up the wrong tree? Is serverless bokeh something that you consider a viable setup or is it more of a curiosity?
>
> --
> You received this message because you are subscribed to the Google Groups "Bokeh Discussion - Public" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to bokeh+un...@continuum.io.
> To post to this group, send email to bo...@continuum.io.
> To view this discussion on the web visit https://groups.google.com/a/continuum.io/d/msgid/bokeh/4284d9ba-5a4b-469a-9bd5-3c4eba2eaf02%40continuum.io\.
> For more options, visit https://groups.google.com/a/continuum.io/d/optout\.

--
You received this message because you are subscribed to the Google Groups "Bokeh Discussion - Public" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [email protected].
To post to this group, send email to [email protected].
To view this discussion on the web visit https://groups.google.com/a/continuum.io/d/msgid/bokeh/54945366-f9a7-48b6-832b-32d5bf146f01%40continuum.io\.
For more options, visit https://groups.google.com/a/continuum.io/d/optout\.

Andy_Baker · April 18, 2016, 9:34am

Apologies for not replying to Bryan’s request for elaboration. This is a side project that I’m only dipping into intermittently.

I’ve rather lost track of what things I struggled with but I’ll try and recall my issues in a general sense:

It wasn’t clear what the CDN url was for the widgets library.
The fact that widgets needed a separate js include wasn’t documented.
The distinction between plots, charts and figures wasn’t clear.
It appears that controlling high-level charts from widgets isn’t currently possible (there’s no source attribute available). Considering the docs guide you initially towards high-level charts then switch to talking about the lower-level level APIs when discussing interactivity threw me off the scent for a while.

Is there an intention to support interactions in the high-level APIs? Although the figure API is fairly simple it doesn’t seem to offer a simple way to create bar charts without manually drawing rectangles. And implementing maps via the lower-level APIs looks fairly tricky. Other chart types look simpler but there’s still a leap between charts and low-level plots that you’re forced to make if you want to integrate widgets.

My dream API would consist of declaring data, charts, layouts and widgets in Python and for most simple cases - describing how the widgets affect the chart also within Python (maybe via some short js callback declarations). Something like this:

plot = Bar(
    data=data,
    label='level',
    values='items',
    group='type',
    legend='top_right',
)

callback = CustomJs(args=dict(bar=plot), code="""
        var data = source.get('data');

        data.filter(data['items'], cb_obj.get('value'));

        source.trigger('change');
    """)

slider = Slider(start=0.1, end=4, value=1, step=.1, title="power", callback=callback)
layout = vform(slider, plot)
bokeh_script, bokeh_html = components(layout)
return TemplateResponse(
    request,
    'bar_chart.html',
    {'bokeh_script': bokeh_script, 'bokeh_html': [bokeh_html]},

···

On Saturday, 2 April 2016 22:27:17 UTC+1, Bryan Van de ven wrote:

Hi Andy,

can you elaborate on:

I was so close but the docs led me astray.

This is an especially important features, so if there is something misleading or missing about the docs, we would want to improve them.

Thanks,

Bryan

On Mar 31, 2016, at 8:14 AM, Andy Baker [email protected] wrote:

Marvellous. Even the callbacks work.

I was so close but the docs led me astray. All I needed was:

def bar_chart(request):
bokeh_script, charts = components(Button(label='hello'))
return TemplateResponse(
    request,
    'bar_chart.html',
    {'bokeh_script': bokeh_script, 'charts': charts,}
)
and this in my template:

On Wednesday, 30 March 2016 15:16:40 UTC+1, [email protected] wrote:

This NYU project uses Bokeh plots served as part of a cherrypy web application:

Here’s the app route that’s serving the plot:

https://github.com/ViDA-NYU/domain_discovery_tool/blob/canavandl/domain_xfilter/vis/server.py#L375

Here’s the template that it’s rendering into:

https://github.com/ViDA-NYU/domain_discovery_tool/blob/canavandl/domain_xfilter/vis/html/cross_filter.html

Looking at your code. I wouldn’t include the CDN in your components method call (I think that’s a deprecated API and doesn’t do anything). Then you’ll have to make sure that you’re loading bokeh-widgets resources (JS and CSS) as a style and script in your template, either from a local directory or CDN.

Resources docs:

http://bokeh.pydata.org/en/latest/docs/user_guide/embed.html#components

As for the error message you’re seeing, it happens when you have a bokeh component that doesn’t have a corresponding BokehJS component. It sometimes happens if you have a mismatch between Bokeh and BokehJS versions, so that might be something to check.

Let us know if that doesn’t work.

lukec

On Wednesday, March 30, 2016 at 3:09:42 AM UTC-5, Andy Baker wrote:

In that case I’m doing something stupid. Something like this:

def widget_test(request):
charts = [Button(label="hello")]
scripts, charts = components(charts, CDN)
return TemplateResponse(
    request,
    'bar_chart.html', {"scripts": scripts, "charts": charts,},
)
(which works perfectly when I swap in a Chart for a button) gives:

Module `Button’ does not exists. The problem may be two fold. Either a model was requested that’s available in an extra bundle, e.g. a widget, or a custom model was requested, but it wasn’t registered before first usage.’

I’ve tried loading bokeh_widgets.js - this gives a different error.

And I’ve tried adding the button to various layout elements instead of passing it directly to components() and I get a similar error.

Is there any example code I could see?

On Monday, 28 March 2016 17:29:37 UTC+1, [email protected] wrote:

It’s entirely possible to use Bokeh widgets in a traditional web framework. Generating the widgets is pretty straightforward, the only additional work is to add the interactivity that you want the widgets to perform. There are several Bokeh-native ways to add callbacks (http://bokeh.pydata.org/en/latest/docs/user_guide/interaction.html#javascript-callbacks) or you can write your own client-side javascript and on-change listener.

On Monday, March 28, 2016 at 6:53:05 AM UTC-5, Andy Baker wrote:

Hi,

I’m using bokeh.embed.components to generate bokeh plots inside a Django app. I’m really keen to avoid using bokeh server due to the increase in deployment and maintenance complexity.

I understand that Bokeh has limited functionality without the server but the docs aren’t terribly clear on exactly what is and isn’t possible. I wanted to generate some widgets and then figure out how to control a client-side plot using them. It seems tantalizingly close to being possible.

Am I barking up the wrong tree? Is serverless bokeh something that you consider a viable setup or is it more of a curiosity?

–
You received this message because you are subscribed to the Google Groups “Bokeh Discussion - Public” group.

To unsubscribe from this group and stop receiving emails from it, send an email to [email protected].

To post to this group, send email to [email protected].

To view this discussion on the web visit https://groups.google.com/a/continuum.io/d/msgid/bokeh/4284d9ba-5a4b-469a-9bd5-3c4eba2eaf02%40continuum.io.

For more options, visit https://groups.google.com/a/continuum.io/d/optout.

Bryan · April 18, 2016, 3:28pm

Hi Andy,

Thanks for the feedback. The "widgets split" is often handled automatically (by output_file, etc.), which is probably why we forgot to say much about it in the docs. But it is important when using e.g. bokeh.embed.components, as you have discovered, so we should should definitely add information about it. I've made an issue here:

Need docs support for BokehJS widgets split · Issue #4186 · bokeh/bokeh · GitHub

Regarding "plots, charts, and figures" one reason for not drawing a big distinction is that ultimately everything is a Plot. Lowest-level Bokeh (bokeh.models) is just a box full of building blocks and no help. Then bokeh.plotting and bokeh.charts are about taking those same set of building blocks, and helping to assemble them in sensible ways automatically. If you have time (no worries if you don't), can you read over this UG section:

http://bokeh.pydata.org/en/latest/docs/user_guide/concepts.html#interfaces

It would be very helpful to know if that clears up questions about how the different pieces fit together. If so, maybe we need to make this section more prominent or accessible. If not, feedback can help expand or refine it.

Regarding high level charts and interactivity, the split-language aspect of Bokeh makes it a bit of a quandry. Right now Charts may do all kinds of grouping and aggregating (in python) and then converts everything to the lower level building blocks I mentioned earlier. This keeps BokehJS smaller, and simpler than if we implemented a high level charting API in JS as well. But then the data for the low level building blocks is "removed" from the original data to the chart, and this is why supporting some kinds of interactivity is problematic. That said, we are definitely interested in supporting what we can, but I would describe this as a more medium term project.

I will mention that I would like to add "vbar" and "hbar" glyphs, so that creating bar-like plots would be simpler from any API. (The "bar" glyphs would be a hybrid of rect and quad: they would take e.g. the bottom and the top in the y direction, but the center and width in the x direction). That might also be helpful, but I don't know exactly when we might be able to get to it. You can follow that issue here:

Add VBar and HBar glyphs · Issue #3423 · bokeh/bokeh · GitHub

Thanks,

Bryan

···

On Apr 18, 2016, at 4:34 AM, Andy Baker <[email protected]> wrote:

------=_Part_7876_655449962.1460972087549
Content-Type: multipart/alternative;
  boundary="----=_Part_7877_1412892260.1460972087550"

------=_Part_7877_1412892260.1460972087550
Content-Type: text/plain; charset=UTF-8

Apologies for not replying to Bryan's request for elaboration. This is a
side project that I'm only dipping into intermittently.

I've rather lost track of what things I struggled with but I'll try and
recall my issues in a general sense:

1. It wasn't clear what the CDN url was for the widgets library.
2. The fact that widgets needed a separate js include wasn't documented.
3. The distinction between plots, charts and figures wasn't clear.
4. It appears that controlling high-level charts from widgets isn't
currently possible (there's no source attribute available). Considering the
docs guide you initially towards high-level charts then switch to talking
about the lower-level level APIs when discussing interactivity threw me off
the scent for a while.

Is there an intention to support interactions in the high-level APIs?
Although the figure API is fairly simple it doesn't seem to offer a simple
way to create bar charts without manually drawing rectangles. And
implementing maps via the lower-level APIs looks fairly tricky. Other chart
types look simpler but there's still a leap between charts and low-level
plots that you're forced to make if you want to integrate widgets.

My dream API would consist of declaring data, charts, layouts and widgets
in Python and for most simple cases - describing how the widgets affect the
chart also within Python (maybe via some short js callback declarations).
Something like this:

plot = Bar(
   data=data,
   label='level',
   values='items',
   group='type',
   legend='top_right',
)

callback = CustomJs(args=dict(bar=plot), code="""
       var data = source.get('data');

       data.filter(data['items'], cb_obj.get('value'));

       source.trigger('change');
   """)

slider = Slider(start=0.1, end=4, value=1, step=.1, title="power", callback=callback)
layout = vform(slider, plot)
bokeh_script, bokeh_html = components(layout)
return TemplateResponse(
   request,
   'bar_chart.html',
   {'bokeh_script': bokeh_script, 'bokeh_html': [bokeh_html]},

)

On Saturday, 2 April 2016 22:27:17 UTC+1, Bryan Van de ven wrote:

Hi Andy,

can you elaborate on:

I was so close but the docs led me astray.

This is an especially important features, so if there is something
misleading or missing about the docs, we would want to improve them.

Thanks,

Bryan

On Mar 31, 2016, at 8:14 AM, Andy Baker <and...@gmail.com <javascript:>> >> wrote:

Marvellous. Even the callbacks work.

I was so close but the docs led me astray. All I needed was:

def bar_chart(request):

   bokeh_script, charts = components(Button(label='hello'))

   return TemplateResponse(
       request,
       'bar_chart.html',
       {'bokeh_script': bokeh_script, 'charts': charts,}
   )

and this in my template:

<link href="http://cdn.pydata.org/bokeh/release/bokeh-0.11.0.css"

rel="stylesheet">

<link href="http://cdn.pydata.org/bokeh/release/bokeh-widgets-0.11.0.css"

rel="stylesheet">

<script src="http://cdn.pydata.org/bokeh/release/bokeh-0.11.0.js"></script>

<script src="http://cdn.pydata.org/bokeh/release/bokeh-widgets-0.11.0.js"></script>

On Wednesday, 30 March 2016 15:16:40 UTC+1, lcan...@continuum.io wrote:
This NYU project uses Bokeh plots served as part of a cherrypy web

application:

Here's the app route that's serving the plot:

https://github.com/ViDA-NYU/domain_discovery_tool/blob/canavandl/domain_xfilter/vis/server.py#L375

Here's the template that it's rendering into:

https://github.com/ViDA-NYU/domain_discovery_tool/blob/canavandl/domain_xfilter/vis/html/cross_filter.html

Looking at your code. I wouldn't include the CDN in your `components`

method call (I think that's a deprecated API and doesn't do anything). Then
you'll have to make sure that you're loading bokeh-widgets resources (JS
and CSS) as a style and script in your template, either from a local
directory or CDN.

Resources docs:
http://bokeh.pydata.org/en/latest/docs/user_guide/embed.html#components

As for the error message you're seeing, it happens when you have a bokeh

component that doesn't have a corresponding BokehJS component. It sometimes
happens if you have a mismatch between Bokeh and BokehJS versions, so that
might be something to check.

Let us know if that doesn't work.

lukec

On Wednesday, March 30, 2016 at 3:09:42 AM UTC-5, Andy Baker wrote:
In that case I'm doing something stupid. Something like this:

def widget_test(request):

   charts = [Button(label="hello")]

   scripts, charts = components(charts, CDN)

   return TemplateResponse(
       request,
       'bar_chart.html', {"scripts": scripts, "charts": charts,},
   )

(which works perfectly when I swap in a Chart for a button) gives:

  Module `Button' does not exists. The problem may be two fold. Either

a model was requested that's available in an extra bundle, e.g. a widget,
or a custom model was requested, but it wasn't registered before first
usage.'

I've tried loading bokeh_widgets.js - this gives a different error.

And I've tried adding the button to various layout elements instead of

passing it directly to components() and I get a similar error.

Is there any example code I could see?

On Monday, 28 March 2016 17:29:37 UTC+1, lcan...@continuum.io wrote:
It's entirely possible to use Bokeh widgets in a traditional web

framework. Generating the widgets is pretty straightforward, the only
additional work is to add the interactivity that you want the widgets to
perform. There are several Bokeh-native ways to add callbacks (
Interaction — Bokeh 3.3.2 Documentation)
or you can write your own client-side javascript and on-change listener.

On Monday, March 28, 2016 at 6:53:05 AM UTC-5, Andy Baker wrote:
Hi,

I'm using bokeh.embed.components to generate bokeh plots inside a Django

app. I'm really keen to avoid using bokeh server due to the increase in
deployment and maintenance complexity.

I understand that Bokeh has limited functionality without the server but

the docs aren't terribly clear on exactly what is and isn't possible. I
wanted to generate some widgets and then figure out how to control a
client-side plot using them. It seems tantalizingly close to being
possible.

Am I barking up the wrong tree? Is serverless bokeh something that you

consider a viable setup or is it more of a curiosity?

--
You received this message because you are subscribed to the Google

Groups "Bokeh Discussion - Public" group.

To unsubscribe from this group and stop receiving emails from it, send

an email to bokeh+un...@continuum.io <javascript:>.

To post to this group, send email to bo...@continuum.io <javascript:>.
To view this discussion on the web visit

https://groups.google.com/a/continuum.io/d/msgid/bokeh/4284d9ba-5a4b-469a-9bd5-3c4eba2eaf02%40continuum.io\.

For more options, visit

https://groups.google.com/a/continuum.io/d/optout\.

--
You received this message because you are subscribed to the Google Groups "Bokeh Discussion - Public" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [email protected].
To post to this group, send email to [email protected].
To view this discussion on the web visit https://groups.google.com/a/continuum.io/d/msgid/bokeh/183747cf-7cc1-4967-a96f-c972fdd8644d%40continuum.io\.
For more options, visit https://groups.google.com/a/continuum.io/d/optout\.

------=_Part_7877_1412892260.1460972087550
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr">Apologies for not replying to Bryan's request for elab=
oration. This is a side project that I'm only dipping into intermittent=
ly.<div><br></div><div>I've rather lost track of what things I struggle=
d with but I'll try and recall my issues in a general sense:</div><div>=
<br></div><div>1. It wasn't clear what the CDN url was for the widgets =
library.</div><div>2. The fact that widgets needed a separate js include wa=
sn't documented.</div><div>3. The distinction between plots, charts and=
figures wasn't clear.</div><div>4. It appears that controlling high-le=
vel charts from widgets isn't currently possible (there's no source=
attribute available). Considering the docs guide you initially towards hig=
h-level charts then switch to talking about the lower-level level APIs when=
discussing interactivity threw me off the scent for a while.</div><div><br=

</div><div>Is there an intention to support interactions in the high-level=

APIs? Although the figure API is fairly simple it doesn't seem to offe=
r a simple way to create bar charts without manually drawing rectangles. An=
d implementing maps via the lower-level APIs looks fairly tricky. Other cha=
rt types look simpler but there's still a leap between charts and low-l=
evel plots that you're forced to make if you want to integrate widgets.=
</div><div><br></div><div>My dream API would consist of declaring data, cha=
rts, layouts and widgets in Python and for most simple cases - describing h=
ow the widgets affect the chart also within Python (maybe via some short js=
callback declarations). Something like this:</div><div><br></div><div><pre=
style=3D"color: rgb(0, 0, 0); font-family: Menlo; font-size: 9pt;">plot =
=3D Bar(<br> <span style=3D"color:#660099;">data</span>=3Ddata,<br> <=
span style=3D"color:#660099;">label</span>=3D<span style=3D"color:#008000;f=
ont-weight:bold;">'level'</span>,<br> <span style=3D"color:#6600=
99;">values</span>=3D<span style=3D"color:#008000;font-weight:bold;">'i=
tems'</span>,<br> <span style=3D"color:#660099;">group</span>=3D<spa=
n style=3D"color:#008000;font-weight:bold;">'type'</span>,<br> <=
span style=3D"color:#660099;">legend</span>=3D<span style=3D"color:#008000;=
font-weight:bold;">'top_right'</span>,<br>)<br><br>callback =3D Cus=
tomJs(<span style=3D"color:#660099;">args</span>=3D<span style=3D"color:#00=
0080;">dict</span>(<span style=3D"color:#660099;">bar</span>=3Dplot), <span=
style=3D"color:#660099;">code</span>=3D<span style=3D"color:#008000;font-w=
eight:bold;">"""<br></span><span style=3D"color:#008000;font=
-weight:bold;"> var data =3D source.get('data');</span><br><=
/pre><pre style=3D"color: rgb(0, 0, 0); font-family: Menlo; font-size: 9pt;=
"><span style=3D"color:#008000;font-weight:bold;"> data.filter(</spa=
><span style=3D"color: rgb(0, 128, 0); font-weight: bold; font-size: 9pt;"=

data['items']</span><span style=3D"font-size: 9pt; color: rgb(0, 1=

28, 0); font-weight: bold;">,=C2=A0</span><span style=3D"font-size: 9pt; co=
lor: rgb(0, 128, 0); font-weight: bold;">cb_obj.get('value')</span>=
<span style=3D"font-size: 9pt; color: rgb(0, 128, 0); font-weight: bold;">)=
;</span></pre><pre style=3D"color: rgb(0, 0, 0); font-family: Menlo; font-s=
ize: 9pt;"><span style=3D"color:#008000;font-weight:bold;"> source.t=
rigger('change');<br></span><span style=3D"color:#008000;font-weigh=
t:bold;"> """</span>)<br><br>slider =3D Slider(<span styl=
e=3D"color:#660099;">start</span>=3D<span style=3D"color:#0000ff;">0.1</spa=
>, <span style=3D"color:#660099;">end</span>=3D<span style=3D"color:#0000f=
f;">4</span>, <span style=3D"color:#660099;">value</span>=3D<span style=3D"=
color:#0000ff;">1</span>, <span style=3D"color:#660099;">step</span>=3D<spa=
n style=3D"color:#0000ff;">.1</span>, <span style=3D"color:#660099;">title<=
/span>=3D<span style=3D"color:#008000;font-weight:bold;">"power"<=
/span>, <span style=3D"color:#660099;">callback</span>=3Dcallback)<br>layou=
t =3D vform(slider, plot)<br>bokeh_script, bokeh_html =3D components(layout=
)<br><span style=3D"color:#000080;font-weight:bold;">return </span>Template=
Response(<br> request,<br> <span style=3D"color:#008000;font-weight:b=
old;">'bar_chart.html'</span>,<br> {'<span style=3D"color:#0=
08000;font-weight:bold;">bokeh_script'</span>: bokeh_script, <span styl=
e=3D"color:#008000;font-weight:bold;">'</span><span style=3D"font-size:=
9pt;">bokeh_html</span><span style=3D"font-size: 9pt; color: rgb(0, 128, 0=
); font-weight: bold;">'</span><span style=3D"font-size: 9pt;">: [</spa=
><span style=3D"font-size: 9pt;">bokeh_html</span><span style=3D"font-size=
: 9pt;">]},</span><br></pre><pre style=3D"color: rgb(0, 0, 0); font-family:=
Menlo; font-size: 9pt;">)</pre><br>On Saturday, 2 April 2016 22:27:17 UTC+=
1, Bryan Van de ven wrote:<blockquote class=3D"gmail_quote" style=3D"margi=
n: 0;margin-left: 0.8ex;border-left: 1px #ccc solid;padding-left: 1ex;">Hi =
Andy,=20
<br>
<br>can you elaborate on:
<br>
<br>> I was so close but the docs led me astray.=20
<br>
<br>This is an especially important features, so if there is something misl=
eading or missing about the docs, we would want to improve them.
<br>
<br>Thanks,
<br>
<br>Bryan
<br>
<br>
<br>
<br>> On Mar 31, 2016, at 8:14 AM, Andy Baker <<a href=3D"javascript:=
" target=3D"_blank" gdf-obfuscated-mailto=3D"zga0qOlNHAAJ" rel=3D"nofollow"=
onmousedown=3D"this.href=3D'javascript:';return true;" onclick=3D"=
this.href=3D'javascript:';return true;">[email protected]</a>> wr=
ote:
<br>>=20
<br>> Marvellous. Even the callbacks work.
<br>>=20
<br>> I was so close but the docs led me astray. All I needed was:
<br>>=20
<br>> def bar_chart(request):
<br>> =C2=A0 =C2=A0=20
<br>> =C2=A0 =C2=A0 bokeh_script, charts =3D components(Button(label=3D&=
#39;<wbr>hello'))
<br>> =C2=A0 =C2=A0=20
<br>> =C2=A0 =C2=A0 return TemplateResponse(
<br>> =C2=A0 =C2=A0 =C2=A0 =C2=A0 request,
<br>> =C2=A0 =C2=A0 =C2=A0 =C2=A0 'bar_chart.html',
<br>> =C2=A0 =C2=A0 =C2=A0 =C2=A0 {'bokeh_script': bokeh_script,=
'charts': charts,}
<br>> =C2=A0 =C2=A0 )
<br>>=20
<br>> and this in my template:
<br>>=20
<br>> <link href=3D"<a href=3D"http://cdn.pydata.org/bokeh/relea=
se/bokeh-0.11.0.css" target=3D"_blank" rel=3D"nofollow" onmousedown=3D"this=
.href=3D'Redirect Notice
keh%2Frelease%2Fbokeh-0.11.0.css\46sa\75D\46sntz\0751\46usg\75AFQjCNE-K8n3s=
6O2dwU748m894V2qRTFEQ';return true;" onclick=3D"this.href=3D'http:/=
/www.google.com/url?q\75http%3A%2F%2Fcdn.pydata.org%2Fbokeh%2Frelease%2Fbok=
eh-0.11.0.css\46sa\75D\46sntz\0751\46usg\75AFQjCNE-K8n3s6O2dwU748m894V2qRTF=
EQ';return true;">http://cdn.pydata.org/<wbr>bokeh/release/bokeh\-0\.11\.0=
.css</a><wbr>" rel=3D"stylesheet">
<br>> <link href=3D"<a href=3D"http://cdn.pydata.org/bokeh/relea=
se/bokeh-widgets-0.11.0.css" target=3D"_blank" rel=3D"nofollow" onmousedown=
=3D"this.href=3D'Redirect Notice
org%2Fbokeh%2Frelease%2Fbokeh-widgets-0.11.0.css\46sa\75D\46sntz\0751\46usg=
\75AFQjCNEVv6qn4dX1HVGR-pAaT0E_fBH8jQ';return true;" onclick=3D"this.hr=
ef=3D'Redirect Notice
%2Frelease%2Fbokeh-widgets-0.11.0.css\46sa\75D\46sntz\0751\46usg\75AFQjCNEV=
v6qn4dX1HVGR-pAaT0E_fBH8jQ';return true;">http://cdn.pydata.org/<wbr>bo=
keh/release/bokeh-widgets-0.<wbr>11.0.css</a>" rel=3D"stylesheet&=
quot;>
<br>> <script src=3D"<a href=3D"http://cdn.pydata.org/bokeh/rele=
ase/bokeh-0.11.0.js" target=3D"_blank" rel=3D"nofollow" onmousedown=3D"this=
.href=3D'Redirect Notice
keh%2Frelease%2Fbokeh-0.11.0.js\46sa\75D\46sntz\0751\46usg\75AFQjCNFQn1sTAc=
TivwBdovqPP7gipZF-Wg';return true;" onclick=3D"this.href=3D'http://=
Redirect Notice\75http%3A%2F%2Fcdn.pydata.org%2Fbokeh%2Frelease%2Fboke=
h-0.11.0.js\46sa\75D\46sntz\0751\46usg\75AFQjCNFQn1sTAcTivwBdovqPP7gipZF-Wg=
';return true;">http://cdn.pydata.org/<wbr>bokeh/release/bokeh\-0\.11\.0\.j=
s</a>"<wbr>></script>
<br>> <script src=3D"<a href=3D"http://cdn.pydata.org/bokeh/rele=
ase/bokeh-widgets-0.11.0.js" target=3D"_blank" rel=3D"nofollow" onmousedown=
=3D"this.href=3D'Redirect Notice
org%2Fbokeh%2Frelease%2Fbokeh-widgets-0.

Barnabas_Szabolcs · April 20, 2016, 9:06pm

Thanks Bryan for the info!
It works now. I did an enhanced version for the components() function where you can structure your
models with any dicts or lists, arbitrarily nested and it will return the divs in the same structure.
I needed this for a page with multiple filters and plots with complex interaction.
I ask my teamlead and if the company allows it we offer to contribute.

Sorry to get back so late, but for me here the discussions get just too much convoluted.
Google groups looks to me far inferior than stackoverflow-ish sites.
In my practice I commonly refer to pandas examples on stackoverflow and using the docs, in 50/50 ratio.
I saw you were reluctant to “expose bokeh complaints” on SO and I would feel the same in your place.
But from my newbie user’s point of view it is far more encouraging to see that issues get solved.
SO serves as a huge cookbook to me. So I would encourage you to have more exposure on SO.
Bokeh is a young but awesome project! Awesome job!

Best,
Barney

···

On Monday, 28 March 2016 13:53:05 UTC+2, Andy Baker wrote:

Hi,

I’m using bokeh.embed.components to generate bokeh plots inside a Django app. I’m really keen to avoid using bokeh server due to the increase in deployment and maintenance complexity.

I understand that Bokeh has limited functionality without the server but the docs aren’t terribly clear on exactly what is and isn’t possible. I wanted to generate some widgets and then figure out how to control a client-side plot using them. It seems tantalizingly close to being possible.

Am I barking up the wrong tree? Is serverless bokeh something that you consider a viable setup or is it more of a curiosity?

Bryan · April 21, 2016, 8:12pm

Hi Barney,

I am glad you got it working.

Regarding SO, my particular issue is that when the answer is already in our docs, and I just post an SO answer that links to our docs (which is all I have the time to do, if the answer is already in the docs I can't afford to spend time to type it up again for no reason) then I get yelled at by people on SO for posting "bad answers".

Bryan

···

On Apr 20, 2016, at 4:06 PM, [email protected] wrote:

Thanks Bryan for the info!
It works now. I did an enhanced version for the components() function where you can structure your
models with any dicts or lists, arbitrarily nested and it will return the divs in the same structure.
I needed this for a page with multiple filters and plots with complex interaction.
I ask my teamlead and if the company allows it we offer to contribute.

Sorry to get back so late, but for me here the discussions get just too much convoluted.
Google groups looks to me far inferior than stackoverflow-ish sites.
In my practice I commonly refer to pandas examples on stackoverflow and using the docs, in 50/50 ratio.
I saw you were reluctant to "expose bokeh complaints" on SO and I would feel the same in your place.
But from my newbie user's point of view it is far more encouraging to see that issues get solved.
SO serves as a huge cookbook to me. So I would encourage you to have more exposure on SO.
Bokeh is a young but awesome project! Awesome job!

Best,
Barney

On Monday, 28 March 2016 13:53:05 UTC+2, Andy Baker wrote:
Hi,

I'm using bokeh.embed.components to generate bokeh plots inside a Django app. I'm really keen to avoid using bokeh server due to the increase in deployment and maintenance complexity.

I understand that Bokeh has limited functionality without the server but the docs aren't terribly clear on exactly what is and isn't possible. I wanted to generate some widgets and then figure out how to control a client-side plot using them. It seems tantalizingly close to being possible.

Am I barking up the wrong tree? Is serverless bokeh something that you consider a viable setup or is it more of a curiosity?

--
You received this message because you are subscribed to the Google Groups "Bokeh Discussion - Public" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [email protected].
To post to this group, send email to [email protected].
To view this discussion on the web visit https://groups.google.com/a/continuum.io/d/msgid/bokeh/bfcab489-b6a5-4215-b5e4-0aeb0ac08f50%40continuum.io\.
For more options, visit https://groups.google.com/a/continuum.io/d/optout\.

Barnabas_Szabolcs · April 22, 2016, 5:04pm

Because they can be assholes! :@@ XD
I see your point, I actually quit the meta community, for a similar reason.
And it is also a nice example why the fb only has up…

If you post a copy-paste quotation and then the link, I think that is generally accepted, a plain link not because other pages (not bokeh) are changing a lot and links go dead very soon.

Barney

···

2016-04-21 22:12 GMT+02:00 Bryan Van de Ven [email protected]:

Hi Barney,

I am glad you got it working.

Regarding SO, my particular issue is that when the answer is already in our docs, and I just post an SO answer that links to our docs (which is all I have the time to do, if the answer is already in the docs I can’t afford to spend time to type it up again for no reason) then I get yelled at by people on SO for posting “bad answers”.

Bryan

On Apr 20, 2016, at 4:06 PM, [email protected] wrote:

Thanks Bryan for the info!

It works now. I did an enhanced version for the components() function where you can structure your

models with any dicts or lists, arbitrarily nested and it will return the divs in the same structure.

I needed this for a page with multiple filters and plots with complex interaction.

I ask my teamlead and if the company allows it we offer to contribute.

Sorry to get back so late, but for me here the discussions get just too much convoluted.

Google groups looks to me far inferior than stackoverflow-ish sites.

In my practice I commonly refer to pandas examples on stackoverflow and using the docs, in 50/50 ratio.

I saw you were reluctant to “expose bokeh complaints” on SO and I would feel the same in your place.

But from my newbie user’s point of view it is far more encouraging to see that issues get solved.

SO serves as a huge cookbook to me. So I would encourage you to have more exposure on SO.

Bokeh is a young but awesome project! Awesome job!

Best,

Barney

On Monday, 28 March 2016 13:53:05 UTC+2, Andy Baker wrote:

Hi,

I’m using bokeh.embed.components to generate bokeh plots inside a Django app. I’m really keen to avoid using bokeh server due to the increase in deployment and maintenance complexity.

I understand that Bokeh has limited functionality without the server but the docs aren’t terribly clear on exactly what is and isn’t possible. I wanted to generate some widgets and then figure out how to control a client-side plot using them. It seems tantalizingly close to being possible.

Am I barking up the wrong tree? Is serverless bokeh something that you consider a viable setup or is it more of a curiosity?

–

You received this message because you are subscribed to the Google Groups “Bokeh Discussion - Public” group.

To unsubscribe from this group and stop receiving emails from it, send an email to [email protected].

To post to this group, send email to [email protected].

To view this discussion on the web visit https://groups.google.com/a/continuum.io/d/msgid/bokeh/bfcab489-b6a5-4215-b5e4-0aeb0ac08f50%40continuum.io.

For more options, visit https://groups.google.com/a/continuum.io/d/optout.

–

You received this message because you are subscribed to a topic in the Google Groups “Bokeh Discussion - Public” group.

To unsubscribe from this topic, visit https://groups.google.com/a/continuum.io/d/topic/bokeh/EesgatXU_2E/unsubscribe.

To unsubscribe from this group and all its topics, send an email to [email protected].

To post to this group, send email to [email protected].

To view this discussion on the web visit https://groups.google.com/a/continuum.io/d/msgid/bokeh/D82AEB9F-AEED-4CAE-ADFA-E5B3859971D7%40continuum.io.
For more options, visit https://groups.google.com/a/continuum.io/d/optout.

Mi a kedvenc kifejezésed? translatian.org