Getting bokeh-server to work with down-sampling

Hello,

I’m migrating my SO question:

http://stackoverflow.com/questions/28099530/bokeh-with-bokeh-server-updating-a-plot-as-it-is-zoomed/28158547

here at the suggestion of someone I think is a member of the Bokeh community (bigreddot). My problem is in trying to use Bokeh’s down-sampling functionality, as demonstrated by Hugo Shi at https://www.youtube.com/watch?v=BinSZ1LKroI

I have an HDF5 file, ‘test.hdf5’, containing a dataset, ‘test’, with two columns, (‘x’ and ‘y’). There are up to 2.5 million pairs of data points. I’m running bokeh-server on port 5006 with a redis backend and injecting Bokeh’s tag into a template with Django (my view code is in the above SO question). However, I don’t see a plot and get a JavaScript Error: “`Error: Domains other than x not supported yet”. I’m not sure what this means.

`

`I managed to get the “animated” demonstration of bokeh-server working, so there’s nothing wrong with my install, but this demonstration doesn’t use downsampling or HDF5, so I don’t have much example code to base my attempt on.

`

One further question: do I have to use HDF5 to implement down-sampling with bokeh-server at the moment, or can I use a memory-mapped file?

Any help would be gratefully received,

Christian

Update: I can get my browser to show the Bokeh Plot window (by adding the argument domain=‘x’ to the line_downsample.source call), but there’s no plot in it. Instead I get a KeyError exception from bokeh-server:

File “/Users/me/anaconda/lib/python2.7/site-packages/bokeh/server/views/data.py”, line 45, in get_data
serverdatasource = clientdoc._models[datasourceid]
KeyError: u’9e502f0b-8266-4d49-b4ab-852d0e429e45’

Can anyone shed any light on this? The docs don’t make it clear how to use HDF5 files with line_downsample.

Christian

···

On 27 January 2015 at 09:55, Christian Hill [email protected] wrote:

Hello,

I’m migrating my SO question:

http://stackoverflow.com/questions/28099530/bokeh-with-bokeh-server-updating-a-plot-as-it-is-zoomed/28158547

here at the suggestion of someone I think is a member of the Bokeh community (bigreddot). My problem is in trying to use Bokeh’s down-sampling functionality, as demonstrated by Hugo Shi at https://www.youtube.com/watch?v=BinSZ1LKroI

I have an HDF5 file, ‘test.hdf5’, containing a dataset, ‘test’, with two columns, (‘x’ and ‘y’). There are up to 2.5 million pairs of data points. I’m running bokeh-server on port 5006 with a redis backend and injecting Bokeh’s tag into a template with Django (my view code is in the above SO question). However, I don’t see a plot and get a JavaScript Error: “`Error: Domains other than x not supported yet”. I’m not sure what this means.

`

`I managed to get the “animated” demonstration of bokeh-server working, so there’s nothing wrong with my install, but this demonstration doesn’t use downsampling or HDF5, so I don’t have much example code to base my attempt on.

`

One further question: do I have to use HDF5 to implement down-sampling with bokeh-server at the moment, or can I use a memory-mapped file?

Any help would be gratefully received,

Christian

Hi Christian,

I don’t knwo if this can help you, but I got the example by Hugo Shi working by the following:

install:

pip install simplejson

further you need to start your server with -D

I started the server from the root dir of the attached .rar

bokeh-server -D /

Sorry for the poor explanation, hope it points you in the right direction!

Kind regards

Mads

test.rar (210 KB)

···

Den tirsdag den 27. januar 2015 kl. 15.55.38 UTC+1 skrev xnx:

Hello,

I’m migrating my SO question:

http://stackoverflow.com/questions/28099530/bokeh-with-bokeh-server-updating-a-plot-as-it-is-zoomed/28158547

here at the suggestion of someone I think is a member of the Bokeh community (bigreddot). My problem is in trying to use Bokeh’s down-sampling functionality, as demonstrated by Hugo Shi at https://www.youtube.com/watch?v=BinSZ1LKroI

I have an HDF5 file, ‘test.hdf5’, containing a dataset, ‘test’, with two columns, (‘x’ and ‘y’). There are up to 2.5 million pairs of data points. I’m running bokeh-server on port 5006 with a redis backend and injecting Bokeh’s tag into a template with Django (my view code is in the above SO question). However, I don’t see a plot and get a JavaScript Error: “`Error: Domains other than x not supported yet”. I’m not sure what this means.

`

`I managed to get the “animated” demonstration of bokeh-server working, so there’s nothing wrong with my install, but this demonstration doesn’t use downsampling or HDF5, so I don’t have much example code to base my attempt on.

`

One further question: do I have to use HDF5 to implement down-sampling with bokeh-server at the moment, or can I use a memory-mapped file?

Any help would be gratefully received,

Christian

Thanks, Mads,
Is Hugo Shi’s source available anywhere? I have arraymanagement and simplejson installed, I’ve started bokeh-server in ~/bokehtemp/ and placed my test.hdf5 file in that same directory. But I don’t know how to refer to the data in my Django view: how do I let Bokeh know to plot the columns x and y from the dataset test inside the test.hdf5 file?

I’ve started bokeh-server with bokeh-server --backend=redis -D / but to no avail…

It’s a shame: I’d really like to get this working for our website.

Best wishes,
Christian

···

On 29 January 2015 at 03:49, Mads Berre [email protected] wrote:

Hi Christian,

I don’t knwo if this can help you, but I got the example by Hugo Shi working by the following:

install:

https://github.com/continuumio/ArrayManagement

pip install simplejson

further you need to start your server with -D

I started the server from the root dir of the attached .rar

bokeh-server -D /

Sorry for the poor explanation, hope it points you in the right direction!

Kind regards

Mads

Den tirsdag den 27. januar 2015 kl. 15.55.38 UTC+1 skrev xnx:

Hello,

I’m migrating my SO question:

http://stackoverflow.com/questions/28099530/bokeh-with-bokeh-server-updating-a-plot-as-it-is-zoomed/28158547

here at the suggestion of someone I think is a member of the Bokeh community (bigreddot). My problem is in trying to use Bokeh’s down-sampling functionality, as demonstrated by Hugo Shi at https://www.youtube.com/watch?v=BinSZ1LKroI

I have an HDF5 file, ‘test.hdf5’, containing a dataset, ‘test’, with two columns, (‘x’ and ‘y’). There are up to 2.5 million pairs of data points. I’m running bokeh-server on port 5006 with a redis backend and injecting Bokeh’s tag into a template with Django (my view code is in the above SO question). However, I don’t see a plot and get a JavaScript Error: “`Error: Domains other than x not supported yet”. I’m not sure what this means.

`

`I managed to get the “animated” demonstration of bokeh-server working, so there’s nothing wrong with my install, but this demonstration doesn’t use downsampling or HDF5, so I don’t have much example code to base my attempt on.

`

One further question: do I have to use HDF5 to implement down-sampling with bokeh-server at the moment, or can I use a memory-mapped file?

Any help would be gratefully received,

Christian

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/fe2f9ccb-4550-4ecf-9ef3-5e6499733edb%40continuum.io.

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

Examples/plotting/serversource.py is similar

We cannot expose the data that was demoed in the video because it is proprietary.

Also the interface for big data will be changing slightly to use blaze instead of array management in the next few weeks(on master)

···

On 29 January 2015 at 03:49, Mads Berre [email protected] wrote:

Hi Christian,

I don’t knwo if this can help you, but I got the example by Hugo Shi working by the following:

install:

https://github.com/continuumio/ArrayManagement

pip install simplejson

further you need to start your server with -D

I started the server from the root dir of the attached .rar

bokeh-server -D /

Sorry for the poor explanation, hope it points you in the right direction!

Kind regards

Mads

Den tirsdag den 27. januar 2015 kl. 15.55.38 UTC+1 skrev xnx:

Hello,

I’m migrating my SO question:

http://stackoverflow.com/questions/28099530/bokeh-with-bokeh-server-updating-a-plot-as-it-is-zoomed/28158547

here at the suggestion of someone I think is a member of the Bokeh community (bigreddot). My problem is in trying to use Bokeh’s down-sampling functionality, as demonstrated by Hugo Shi at https://www.youtube.com/watch?v=BinSZ1LKroI

I have an HDF5 file, ‘test.hdf5’, containing a dataset, ‘test’, with two columns, (‘x’ and ‘y’). There are up to 2.5 million pairs of data points. I’m running bokeh-server on port 5006 with a redis backend and injecting Bokeh’s tag into a template with Django (my view code is in the above SO question). However, I don’t see a plot and get a JavaScript Error: “`Error: Domains other than x not supported yet”. I’m not sure what this means.

`

`I managed to get the “animated” demonstration of bokeh-server working, so there’s nothing wrong with my install, but this demonstration doesn’t use downsampling or HDF5, so I don’t have much example code to base my attempt on.

`

One further question: do I have to use HDF5 to implement down-sampling with bokeh-server at the moment, or can I use a memory-mapped file?

Any help would be gratefully received,

Christian

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/fe2f9ccb-4550-4ecf-9ef3-5e6499733edb%40continuum.io.

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

That sound great, I will look forward to a improved version!

@hugo, I have not succeed in producing my own hdf5 file for the array management to read, is there some examples available on how to construct a file that can be used?

@Christian, I found the source code for the example here: http://continuum.io/blog/bokeh-0.4.1 (in the “what’s awesome” section)

This feature is very useful and I hope that there will be some good examples in the upcoming implementation using blaze!

Kind regards
Mads

···

Den fredag den 30. januar 2015 kl. 04.22.27 UTC+1 skrev Hugo Shi:

Examples/plotting/serversource.py is similar

We cannot expose the data that was demoed in the video because it is proprietary.

Also the interface for big data will be changing slightly to use blaze instead of array management in the next few weeks(on master)

On Jan 29, 2015 7:54 PM, “Christian Hill” [email protected] wrote:

Thanks, Mads,
Is Hugo Shi’s source available anywhere? I have arraymanagement and simplejson installed, I’ve started bokeh-server in ~/bokehtemp/ and placed my test.hdf5 file in that same directory. But I don’t know how to refer to the data in my Django view: how do I let Bokeh know to plot the columns x and y from the dataset test inside the test.hdf5 file?

I’ve started bokeh-server with bokeh-server --backend=redis -D / but to no avail…

It’s a shame: I’d really like to get this working for our website.

Best wishes,
Christian

On 29 January 2015 at 03:49, Mads Berre [email protected] wrote:

Hi Christian,

I don’t knwo if this can help you, but I got the example by Hugo Shi working by the following:

install:

https://github.com/continuumio/ArrayManagement

pip install simplejson

further you need to start your server with -D

I started the server from the root dir of the attached .rar

bokeh-server -D /

Sorry for the poor explanation, hope it points you in the right direction!

Kind regards

Mads

Den tirsdag den 27. januar 2015 kl. 15.55.38 UTC+1 skrev xnx:

Hello,

I’m migrating my SO question:

http://stackoverflow.com/questions/28099530/bokeh-with-bokeh-server-updating-a-plot-as-it-is-zoomed/28158547

here at the suggestion of someone I think is a member of the Bokeh community (bigreddot). My problem is in trying to use Bokeh’s down-sampling functionality, as demonstrated by Hugo Shi at https://www.youtube.com/watch?v=BinSZ1LKroI

I have an HDF5 file, ‘test.hdf5’, containing a dataset, ‘test’, with two columns, (‘x’ and ‘y’). There are up to 2.5 million pairs of data points. I’m running bokeh-server on port 5006 with a redis backend and injecting Bokeh’s tag into a template with Django (my view code is in the above SO question). However, I don’t see a plot and get a JavaScript Error: “`Error: Domains other than x not supported yet”. I’m not sure what this means.

`

`I managed to get the “animated” demonstration of bokeh-server working, so there’s nothing wrong with my install, but this demonstration doesn’t use downsampling or HDF5, so I don’t have much example code to base my attempt on.

`

One further question: do I have to use HDF5 to implement down-sampling with bokeh-server at the moment, or can I use a memory-mapped file?

Any help would be gratefully received,

Christian

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/fe2f9ccb-4550-4ecf-9ef3-5e6499733edb%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/CABtdkq77D1QZjR3s91vx6zZWyPciO3JVc3BCtdiUgv5RD%3Dv4aA%40mail.gmail.com.

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

Dear Mads and Hugo,

Thanks for your interest in my problem. I’m afraid I can’t get the remotedata.py example at https://github.com/bokeh/bokeh/blob/demo/examples/plotting/server/remotedata.py to work: this script imports (but does not seem to use) line_downsample, and uses a class, RemoteDataSource, which isn’t in my version of bokeh.plotting (0.7.1, installed with conda).

If I try to implement the Examples/plotting/serversource.py example, I get the same KeyError if I try to embed with
bplot.push()
tag = embed.autoload_server(p, bplot.cursession())

and an AttributeError if I just look on the bokeh-server at 127.0.0.1:5006:

File “/Users/me/anaconda/lib/python2.7/site-packages/arraymanagement/client.py”, line 72, in get_node
node = node.get_node(n)
AttributeError: ‘NoneType’ object has no attribute ‘get_node’

Should I wait until the next update of bokeh to implement this?

Christian

PS I’m a little surprised to discover that Apple’s share price is considered proprietary data :slight_smile:

···

On 30 January 2015 at 01:54, Mads Berre [email protected] wrote:

That sound great, I will look forward to a improved version!

@hugo, I have not succeed in producing my own hdf5 file for the array management to read, is there some examples available on how to construct a file that can be used?

@Christian, I found the source code for the example here: http://continuum.io/blog/bokeh-0.4.1 (in the “what’s awesome” section)

This feature is very useful and I hope that there will be some good examples in the upcoming implementation using blaze!

Kind regards
Mads

Den fredag den 30. januar 2015 kl. 04.22.27 UTC+1 skrev Hugo Shi:

Examples/plotting/serversource.py is similar

We cannot expose the data that was demoed in the video because it is proprietary.

Also the interface for big data will be changing slightly to use blaze instead of array management in the next few weeks(on master)

On Jan 29, 2015 7:54 PM, “Christian Hill” [email protected] wrote:

Thanks, Mads,
Is Hugo Shi’s source available anywhere? I have arraymanagement and simplejson installed, I’ve started bokeh-server in ~/bokehtemp/ and placed my test.hdf5 file in that same directory. But I don’t know how to refer to the data in my Django view: how do I let Bokeh know to plot the columns x and y from the dataset test inside the test.hdf5 file?

I’ve started bokeh-server with bokeh-server --backend=redis -D / but to no avail…

It’s a shame: I’d really like to get this working for our website.

Best wishes,
Christian

On 29 January 2015 at 03:49, Mads Berre [email protected] wrote:

Hi Christian,

I don’t knwo if this can help you, but I got the example by Hugo Shi working by the following:

install:

https://github.com/continuumio/ArrayManagement

pip install simplejson

further you need to start your server with -D

I started the server from the root dir of the attached .rar

bokeh-server -D /

Sorry for the poor explanation, hope it points you in the right direction!

Kind regards

Mads

Den tirsdag den 27. januar 2015 kl. 15.55.38 UTC+1 skrev xnx:

Hello,

I’m migrating my SO question:

http://stackoverflow.com/questions/28099530/bokeh-with-bokeh-server-updating-a-plot-as-it-is-zoomed/28158547

here at the suggestion of someone I think is a member of the Bokeh community (bigreddot). My problem is in trying to use Bokeh’s down-sampling functionality, as demonstrated by Hugo Shi at https://www.youtube.com/watch?v=BinSZ1LKroI

I have an HDF5 file, ‘test.hdf5’, containing a dataset, ‘test’, with two columns, (‘x’ and ‘y’). There are up to 2.5 million pairs of data points. I’m running bokeh-server on port 5006 with a redis backend and injecting Bokeh’s tag into a template with Django (my view code is in the above SO question). However, I don’t see a plot and get a JavaScript Error: “`Error: Domains other than x not supported yet”. I’m not sure what this means.

`

`I managed to get the “animated” demonstration of bokeh-server working, so there’s nothing wrong with my install, but this demonstration doesn’t use downsampling or HDF5, so I don’t have much example code to base my attempt on.

`

One further question: do I have to use HDF5 to implement down-sampling with bokeh-server at the moment, or can I use a memory-mapped file?

Any help would be gratefully received,

Christian

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/fe2f9ccb-4550-4ecf-9ef3-5e6499733edb%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/CABtdkq77D1QZjR3s91vx6zZWyPciO3JVc3BCtdiUgv5RD%3Dv4aA%40mail.gmail.com.

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/484f0018-0503-4c68-9a04-f5c81d0150e4%40continuum.io.

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

Hi
As I remember I have seen this error when I started the bokeh-server without setting the data storage directory with the -D argument. Are you sure that you have set this argument to the server correctly.

I too, haven’t been able to find any information about how to make my own dataset to use for plotting on the bokeh-server with downsampling. I have investigated the format of the hdf5 files, but haven’t been able to find any python package which can help to write such a format. I don’t understand why I am unable to find any documentation of this format at all.

I hope that the -D arguments help you making the examples work.

Kind Regards

Andreas

···

On Friday, 30 January 2015 16:43:46 UTC+1, xnx wrote:

Dear Mads and Hugo,

Thanks for your interest in my problem. I’m afraid I can’t get the remotedata.py example at https://github.com/bokeh/bokeh/blob/demo/examples/plotting/server/remotedata.py to work: this script imports (but does not seem to use) line_downsample, and uses a class, RemoteDataSource, which isn’t in my version of bokeh.plotting (0.7.1, installed with conda).

If I try to implement the Examples/plotting/serversource.py example, I get the same KeyError if I try to embed with
bplot.push()
tag = embed.autoload_server(p, bplot.cursession())

Should I wait until the next update of bokeh to implement this?

Christian

PS I’m a little surprised to discover that Apple’s share price is considered proprietary data :slight_smile:

and an AttributeError if I just look on the bokeh-server at 127.0.0.1:5006:

File “/Users/me/anaconda/lib/python2.7/site-packages/arraymanagement/client.py”, line 72, in get_node
node = node.get_node(n)
AttributeError: ‘NoneType’ object has no attribute ‘get_node’

On 30 January 2015 at 01:54, Mads Berre [email protected] wrote:

That sound great, I will look forward to a improved version!

@hugo, I have not succeed in producing my own hdf5 file for the array management to read, is there some examples available on how to construct a file that can be used?

@Christian, I found the source code for the example here: http://continuum.io/blog/bokeh-0.4.1 (in the “what’s awesome” section)

This feature is very useful and I hope that there will be some good examples in the upcoming implementation using blaze!

Kind regards
Mads

Den fredag den 30. januar 2015 kl. 04.22.27 UTC+1 skrev Hugo Shi:

Examples/plotting/serversource.py is similar

We cannot expose the data that was demoed in the video because it is proprietary.

Also the interface for big data will be changing slightly to use blaze instead of array management in the next few weeks(on master)

On Jan 29, 2015 7:54 PM, “Christian Hill” [email protected] wrote:

Thanks, Mads,
Is Hugo Shi’s source available anywhere? I have arraymanagement and simplejson installed, I’ve started bokeh-server in ~/bokehtemp/ and placed my test.hdf5 file in that same directory. But I don’t know how to refer to the data in my Django view: how do I let Bokeh know to plot the columns x and y from the dataset test inside the test.hdf5 file?

I’ve started bokeh-server with bokeh-server --backend=redis -D / but to no avail…

It’s a shame: I’d really like to get this working for our website.

Best wishes,
Christian

On 29 January 2015 at 03:49, Mads Berre [email protected] wrote:

Hi Christian,

I don’t knwo if this can help you, but I got the example by Hugo Shi working by the following:

install:

https://github.com/continuumio/ArrayManagement

pip install simplejson

further you need to start your server with -D

I started the server from the root dir of the attached .rar

bokeh-server -D /

Sorry for the poor explanation, hope it points you in the right direction!

Kind regards

Mads

Den tirsdag den 27. januar 2015 kl. 15.55.38 UTC+1 skrev xnx:

Hello,

I’m migrating my SO question:

http://stackoverflow.com/questions/28099530/bokeh-with-bokeh-server-updating-a-plot-as-it-is-zoomed/28158547

here at the suggestion of someone I think is a member of the Bokeh community (bigreddot). My problem is in trying to use Bokeh’s down-sampling functionality, as demonstrated by Hugo Shi at https://www.youtube.com/watch?v=BinSZ1LKroI

I have an HDF5 file, ‘test.hdf5’, containing a dataset, ‘test’, with two columns, (‘x’ and ‘y’). There are up to 2.5 million pairs of data points. I’m running bokeh-server on port 5006 with a redis backend and injecting Bokeh’s tag into a template with Django (my view code is in the above SO question). However, I don’t see a plot and get a JavaScript Error: “`Error: Domains other than x not supported yet”. I’m not sure what this means.

`

`I managed to get the “animated” demonstration of bokeh-server working, so there’s nothing wrong with my install, but this demonstration doesn’t use downsampling or HDF5, so I don’t have much example code to base my attempt on.

`

One further question: do I have to use HDF5 to implement down-sampling with bokeh-server at the moment, or can I use a memory-mapped file?

Any help would be gratefully received,

Christian

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/fe2f9ccb-4550-4ecf-9ef3-5e6499733edb%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/CABtdkq77D1QZjR3s91vx6zZWyPciO3JVc3BCtdiUgv5RD%3Dv4aA%40mail.gmail.com.

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/484f0018-0503-4c68-9a04-f5c81d0150e4%40continuum.io.

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

Hi everyone,

I wanted to point everyone at a particular PR:

  https://github.com/bokeh/bokeh/pull/1713

We are working hard to clean up and improve the experience of using the server for downsampled data. Basically to date we had developed capability, but it was difficult to deploy and use. Very hopeful that this PR together with some better docs and example will be merged for our upcoming 0.8 release.

Bryan

···

On Feb 2, 2015, at 1:57 AM, [email protected] wrote:

Hi
As I remember I have seen this error when I started the bokeh-server without setting the data storage directory with the -D argument. Are you sure that you have set this argument to the server correctly.

I too, haven't been able to find any information about how to make my own dataset to use for plotting on the bokeh-server with downsampling. I have investigated the format of the hdf5 files, but haven't been able to find any python package which can help to write such a format. I don't understand why I am unable to find any documentation of this format at all.

I hope that the -D arguments help you making the examples work.

Kind Regards

Andreas

On Friday, 30 January 2015 16:43:46 UTC+1, xnx wrote:
Dear Mads and Hugo,
Thanks for your interest in my problem. I'm afraid I can't get the remotedata.py example at https://github.com/bokeh/bokeh/blob/demo/examples/plotting/server/remotedata.py to work: this script imports (but does not seem to use) line_downsample, and uses a class, RemoteDataSource, which isn't in my version of bokeh.plotting (0.7.1, installed with conda).

If I try to implement the Examples/plotting/serversource.py example, I get the same KeyError if I try to embed with
    bplot.push()
    tag = embed.autoload_server(p, bplot.cursession())

and an AttributeError if I just look on the bokeh-server at 127.0.0.1:5006:

  File "/Users/me/anaconda/lib/python2.7/site-packages/arraymanagement/client.py", line 72, in get_node
    node = node.get_node(n)
AttributeError: 'NoneType' object has no attribute 'get_node'

Should I wait until the next update of bokeh to implement this?
Christian

PS I'm a little surprised to discover that Apple's share price is considered proprietary data :slight_smile:

On 30 January 2015 at 01:54, Mads Berre <[email protected]> wrote:
That sound great, I will look forward to a improved version!

@hugo, I have not succeed in producing my own hdf5 file for the array management to read, is there some examples available on how to construct a file that can be used?

@Christian, I found the source code for the example here: http://continuum.io/blog/bokeh-0.4.1 (in the "what's awesome" section)

This feature is very useful and I hope that there will be some good examples in the upcoming implementation using blaze!

Kind regards
Mads

Den fredag den 30. januar 2015 kl. 04.22.27 UTC+1 skrev Hugo Shi:
Examples/plotting/serversource.py is similar

We cannot expose the data that was demoed in the video because it is proprietary.

Also the interface for big data will be changing slightly to use blaze instead of array management in the next few weeks(on master)

On Jan 29, 2015 7:54 PM, "Christian Hill" <[email protected]> wrote:
Thanks, Mads,
Is Hugo Shi's source available anywhere? I have arraymanagement and simplejson installed, I've started bokeh-server in ~/bokehtemp/ and placed my test.hdf5 file in that same directory. But I don't know how to refer to the data in my Django view: how do I let Bokeh know to plot the columns x and y from the dataset test inside the test.hdf5 file?
I've started bokeh-server with bokeh-server --backend=redis -D / but to no avail...
It's a shame: I'd really like to get this working for our website.
Best wishes,
Christian

On 29 January 2015 at 03:49, Mads Berre <[email protected]> wrote:
Hi Christian,

I don't knwo if this can help you, but I got the example by Hugo Shi working by the following:
install:
https://github.com/continuumio/ArrayManagement
pip install simplejson

further you need to start your server with -D
I started the server from the root dir of the attached .rar
bokeh-server -D /

Sorry for the poor explanation, hope it points you in the right direction!

Kind regards
Mads

Den tirsdag den 27. januar 2015 kl. 15.55.38 UTC+1 skrev xnx:
Hello,
I'm migrating my SO question:

http://stackoverflow.com/questions/28099530/bokeh-with-bokeh-server-updating-a-plot-as-it-is-zoomed/28158547

here at the suggestion of someone I think is a member of the Bokeh community (bigreddot). My problem is in trying to use Bokeh's down-sampling functionality, as demonstrated by Hugo Shi at https://www.youtube.com/watch?v=BinSZ1LKroI

I have an HDF5 file, 'test.hdf5', containing a dataset, 'test', with two columns, ('x' and 'y'). There are up to 2.5 million pairs of data points. I'm running bokeh-server on port 5006 with a redis backend and injecting Bokeh's <script> tag into a template with Django (my view code is in the above SO question). However, I don't see a plot and get a JavaScript Error: "Error: Domains other than x not supported yet". I'm not sure what this means.

I managed to get the "animated" demonstration of bokeh-server working, so there's nothing wrong with my install, but this demonstration doesn't use downsampling or HDF5, so I don't have much example code to base my attempt on.

One further question: do I have to use HDF5 to implement down-sampling with bokeh-server at the moment, or can I use a memory-mapped file?

Any help would be gratefully received,
Christian

--
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/fe2f9ccb-4550-4ecf-9ef3-5e6499733edb%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/CABtdkq77D1QZjR3s91vx6zZWyPciO3JVc3BCtdiUgv5RD%3Dv4aA%40mail.gmail.com.
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/484f0018-0503-4c68-9a04-f5c81d0150e4%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/d7c7e6ca-d886-44ac-b537-c52a7865cc8b%40continuum.io.
For more options, visit https://groups.google.com/a/continuum.io/d/optout.

Sounds great. I’m really looking forward to this.

···

On Monday, 2 February 2015 17:15:24 UTC+1, Bryan Van de ven wrote:

Hi everyone,

I wanted to point everyone at a particular PR:

    [https://github.com/bokeh/bokeh/pull/1713](https://github.com/bokeh/bokeh/pull/1713)

We are working hard to clean up and improve the experience of using the server for downsampled data. Basically to date we had developed capability, but it was difficult to deploy and use. Very hopeful that this PR together with some better docs and example will be merged for our upcoming 0.8 release.

Bryan

On Feb 2, 2015, at 1:57 AM, [email protected] wrote:

Hi

As I remember I have seen this error when I started the bokeh-server without setting the data storage directory with the -D argument. Are you sure that you have set this argument to the server correctly.

I too, haven’t been able to find any information about how to make my own dataset to use for plotting on the bokeh-server with downsampling. I have investigated the format of the hdf5 files, but haven’t been able to find any python package which can help to write such a format. I don’t understand why I am unable to find any documentation of this format at all.

I hope that the -D arguments help you making the examples work.

Kind Regards

Andreas

On Friday, 30 January 2015 16:43:46 UTC+1, xnx wrote:

Dear Mads and Hugo,

Thanks for your interest in my problem. I’m afraid I can’t get the remotedata.py example at https://github.com/bokeh/bokeh/blob/demo/examples/plotting/server/remotedata.py to work: this script imports (but does not seem to use) line_downsample, and uses a class, RemoteDataSource, which isn’t in my version of bokeh.plotting (0.7.1, installed with conda).

If I try to implement the Examples/plotting/serversource.py example, I get the same KeyError if I try to embed with
bplot.push()

tag = embed.autoload_server(p, bplot.cursession())

and an AttributeError if I just look on the bokeh-server at 127.0.0.1:5006:

File “/Users/me/anaconda/lib/python2.7/site-packages/arraymanagement/client.py”, line 72, in get_node

node = node.get_node(n)

AttributeError: ‘NoneType’ object has no attribute ‘get_node’

Should I wait until the next update of bokeh to implement this?

Christian

PS I’m a little surprised to discover that Apple’s share price is considered proprietary data :slight_smile:

On 30 January 2015 at 01:54, Mads Berre [email protected] wrote:

That sound great, I will look forward to a improved version!

@hugo, I have not succeed in producing my own hdf5 file for the array management to read, is there some examples available on how to construct a file that can be used?

@Christian, I found the source code for the example here: http://continuum.io/blog/bokeh-0.4.1 (in the “what’s awesome” section)

This feature is very useful and I hope that there will be some good examples in the upcoming implementation using blaze!

Kind regards

Mads

Den fredag den 30. januar 2015 kl. 04.22.27 UTC+1 skrev Hugo Shi:

Examples/plotting/serversource.py is similar

We cannot expose the data that was demoed in the video because it is proprietary.

Also the interface for big data will be changing slightly to use blaze instead of array management in the next few weeks(on master)

On Jan 29, 2015 7:54 PM, “Christian Hill” [email protected] wrote:

Thanks, Mads,

Is Hugo Shi’s source available anywhere? I have arraymanagement and simplejson installed, I’ve started bokeh-server in ~/bokehtemp/ and placed my test.hdf5 file in that same directory. But I don’t know how to refer to the data in my Django view: how do I let Bokeh know to plot the columns x and y from the dataset test inside the test.hdf5 file?

I’ve started bokeh-server with bokeh-server --backend=redis -D / but to no avail…

It’s a shame: I’d really like to get this working for our website.

Best wishes,

Christian

On 29 January 2015 at 03:49, Mads Berre [email protected] wrote:

Hi Christian,

I don’t knwo if this can help you, but I got the example by Hugo Shi working by the following:

install:
https://github.com/continuumio/ArrayManagement

pip install simplejson

further you need to start your server with -D
I started the server from the root dir of the attached .rar

bokeh-server -D /

Sorry for the poor explanation, hope it points you in the right direction!

Kind regards

Mads

Den tirsdag den 27. januar 2015 kl. 15.55.38 UTC+1 skrev xnx:

Hello,

I’m migrating my SO question:

http://stackoverflow.com/questions/28099530/bokeh-with-bokeh-server-updating-a-plot-as-it-is-zoomed/28158547

here at the suggestion of someone I think is a member of the Bokeh community (bigreddot). My problem is in trying to use Bokeh’s down-sampling functionality, as demonstrated by Hugo Shi at https://www.youtube.com/watch?v=BinSZ1LKroI

I have an HDF5 file, ‘test.hdf5’, containing a dataset, ‘test’, with two columns, (‘x’ and ‘y’). There are up to 2.5 million pairs of data points. I’m running bokeh-server on port 5006 with a redis backend and injecting Bokeh’s tag into a template with Django (my view code is in the above SO question). However, I don’t see a plot and get a JavaScript Error: “Error: Domains other than x not supported yet”. I’m not sure what this means.

I managed to get the “animated” demonstration of bokeh-server working, so there’s nothing wrong with my install, but this demonstration doesn’t use downsampling or HDF5, so I don’t have much example code to base my attempt on.

One further question: do I have to use HDF5 to implement down-sampling with bokeh-server at the moment, or can I use a memory-mapped file?

Any help would be gratefully received,

Christian


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/fe2f9ccb-4550-4ecf-9ef3-5e6499733edb%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/CABtdkq77D1QZjR3s91vx6zZWyPciO3JVc3BCtdiUgv5RD%3Dv4aA%40mail.gmail.com.

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/484f0018-0503-4c68-9a04-f5c81d0150e4%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/d7c7e6ca-d886-44ac-b537-c52a7865cc8b%40continuum.io.

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

Bryan,

Bokeh 0.8+ is available, are the “better docs” now shipped along with it? I too am interested in testing out the downsampling for a scatterplot matrix. Any reproducible example(s) would be greatly appreciated.

Thanks!

Sean

···

On Monday, February 2, 2015 at 11:15:24 AM UTC-5, Bryan Van de ven wrote:

Hi everyone,

I wanted to point everyone at a particular PR:

    [https://github.com/bokeh/bokeh/pull/1713](https://github.com/bokeh/bokeh/pull/1713)

We are working hard to clean up and improve the experience of using the server for downsampled data. Basically to date we had developed capability, but it was difficult to deploy and use. Very hopeful that this PR together with some better docs and example will be merged for our upcoming 0.8 release.

Bryan

On Feb 2, 2015, at 1:57 AM, [email protected] wrote:

Hi

As I remember I have seen this error when I started the bokeh-server without setting the data storage directory with the -D argument. Are you sure that you have set this argument to the server correctly.

I too, haven’t been able to find any information about how to make my own dataset to use for plotting on the bokeh-server with downsampling. I have investigated the format of the hdf5 files, but haven’t been able to find any python package which can help to write such a format. I don’t understand why I am unable to find any documentation of this format at all.

I hope that the -D arguments help you making the examples work.

Kind Regards

Andreas

On Friday, 30 January 2015 16:43:46 UTC+1, xnx wrote:

Dear Mads and Hugo,

Thanks for your interest in my problem. I’m afraid I can’t get the remotedata.py example at https://github.com/bokeh/bokeh/blob/demo/examples/plotting/server/remotedata.py to work: this script imports (but does not seem to use) line_downsample, and uses a class, RemoteDataSource, which isn’t in my version of bokeh.plotting (0.7.1, installed with conda).

If I try to implement the Examples/plotting/serversource.py example, I get the same KeyError if I try to embed with
bplot.push()

tag = embed.autoload_server(p, bplot.cursession())

and an AttributeError if I just look on the bokeh-server at 127.0.0.1:5006:

File “/Users/me/anaconda/lib/python2.7/site-packages/arraymanagement/client.py”, line 72, in get_node

node = node.get_node(n)

AttributeError: ‘NoneType’ object has no attribute ‘get_node’

Should I wait until the next update of bokeh to implement this?

Christian

PS I’m a little surprised to discover that Apple’s share price is considered proprietary data :slight_smile:

On 30 January 2015 at 01:54, Mads Berre [email protected] wrote:

That sound great, I will look forward to a improved version!

@hugo, I have not succeed in producing my own hdf5 file for the array management to read, is there some examples available on how to construct a file that can be used?

@Christian, I found the source code for the example here: http://continuum.io/blog/bokeh-0.4.1 (in the “what’s awesome” section)

This feature is very useful and I hope that there will be some good examples in the upcoming implementation using blaze!

Kind regards

Mads

Den fredag den 30. januar 2015 kl. 04.22.27 UTC+1 skrev Hugo Shi:

Examples/plotting/serversource.py is similar

We cannot expose the data that was demoed in the video because it is proprietary.

Also the interface for big data will be changing slightly to use blaze instead of array management in the next few weeks(on master)

On Jan 29, 2015 7:54 PM, “Christian Hill” [email protected] wrote:

Thanks, Mads,

Is Hugo Shi’s source available anywhere? I have arraymanagement and simplejson installed, I’ve started bokeh-server in ~/bokehtemp/ and placed my test.hdf5 file in that same directory. But I don’t know how to refer to the data in my Django view: how do I let Bokeh know to plot the columns x and y from the dataset test inside the test.hdf5 file?

I’ve started bokeh-server with bokeh-server --backend=redis -D / but to no avail…

It’s a shame: I’d really like to get this working for our website.

Best wishes,

Christian

On 29 January 2015 at 03:49, Mads Berre [email protected] wrote:

Hi Christian,

I don’t knwo if this can help you, but I got the example by Hugo Shi working by the following:

install:
https://github.com/continuumio/ArrayManagement

pip install simplejson

further you need to start your server with -D
I started the server from the root dir of the attached .rar

bokeh-server -D /

Sorry for the poor explanation, hope it points you in the right direction!

Kind regards

Mads

Den tirsdag den 27. januar 2015 kl. 15.55.38 UTC+1 skrev xnx:

Hello,

I’m migrating my SO question:

http://stackoverflow.com/questions/28099530/bokeh-with-bokeh-server-updating-a-plot-as-it-is-zoomed/28158547

here at the suggestion of someone I think is a member of the Bokeh community (bigreddot). My problem is in trying to use Bokeh’s down-sampling functionality, as demonstrated by Hugo Shi at https://www.youtube.com/watch?v=BinSZ1LKroI

I have an HDF5 file, ‘test.hdf5’, containing a dataset, ‘test’, with two columns, (‘x’ and ‘y’). There are up to 2.5 million pairs of data points. I’m running bokeh-server on port 5006 with a redis backend and injecting Bokeh’s tag into a template with Django (my view code is in the above SO question). However, I don’t see a plot and get a JavaScript Error: “Error: Domains other than x not supported yet”. I’m not sure what this means.

I managed to get the “animated” demonstration of bokeh-server working, so there’s nothing wrong with my install, but this demonstration doesn’t use downsampling or HDF5, so I don’t have much example code to base my attempt on.

One further question: do I have to use HDF5 to implement down-sampling with bokeh-server at the moment, or can I use a memory-mapped file?

Any help would be gratefully received,

Christian


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/fe2f9ccb-4550-4ecf-9ef3-5e6499733edb%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/CABtdkq77D1QZjR3s91vx6zZWyPciO3JVc3BCtdiUgv5RD%3Dv4aA%40mail.gmail.com.

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/484f0018-0503-4c68-9a04-f5c81d0150e4%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/d7c7e6ca-d886-44ac-b537-c52a7865cc8b%40continuum.io.

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