WIP real-time base client by teonbrooks · Pull Request #4110 · mne-tools/mne-python

teonbrooks · 2017-03-28T20:09:09Z

TODOs

Add example for LSL client (using simulation)
Add test for _BaseClient
Add tests for LSL client

jasmainak · 2017-03-29T05:26:25Z

@teonbrooks have you looked at this by @alexandrebarachant : https://github.com/alexandrebarachant/muse-lsl ? Let's try to finish the refactoring tomorrow morning and then go for the openbci client later in the day ...

jasmainak · 2017-03-29T05:27:18Z

mne/realtime/client.py

-#          Martin Luessi <mluessi@nmr.mgh.harvard.edu>
-#          Matti Hamalainen <msh@nmr.mgh.harvard.edu>
+# Authors: Teon Brooks <teon.brooks@gmail.com>
+#          Mainak Jas <mainak@neuro.hut.fi>


I think the diff is really unreadable if we rename the files. I'd prefer keeping the filenames as is for now ...

codecov-io · 2017-03-29T18:13:19Z

Codecov Report

Merging #4110 into master will decrease coverage by 0.23%.
The diff coverage is 31.57%.

@@            Coverage Diff             @@
##           master    #4110      +/-   ##
==========================================
- Coverage   86.18%   85.95%   -0.24%     
==========================================
  Files         354      355       +1     
  Lines       63729    63755      +26     
  Branches     9709     9709              
==========================================
- Hits        54927    54801     -126     
- Misses       6127     6272     +145     
- Partials     2675     2682       +7

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update a092868...fe46b97. Read the comment docs.

rodrigohubner · 2017-04-02T03:39:26Z

I'll try to participate in this WIP (+1).

jasmainak · 2017-04-02T05:02:15Z

Great! Feel free to review it, pull the changes and try them on your computer

rodrigohubner

Two doubts:

For me to get the whole project should I pull in Teon's repository?
How would I contribute to commit to this same PR?

rodrigohubner · 2017-04-05T01:58:15Z

mne/realtime/base_client.py

 # License: BSD (3-clause)

-import copy
+import threading


I think it is better to work with multiprocessing because of known problems with multithreading in python (GIL for example).

if you have a better solution, i welcome it. here we are just extracting away from our existing code and wanted to have a lightweight base class object that would be work across our existing clients. I wasn't aware of the problems with multithreading

I used thread for similar function and never had any issues. Yes python has problem with the GIL, but that does not mean thread should never be used. What you want to avoid is having a thread that never release the GIL (i.e. continuously run). For example, when using LSL.pull_sample() or a socket.recv, those function will block until you receive data but this is made with a call to a C extension or to a shared library that will effectively release the GIL during the time it wait.

Thanks @alexandrebarachant . I'd say let's stick to threading for now, get the API right, add an example, so people can pull and test. Then we can smooth out other details. Make it work, then make it nice :)

jaeilepp · 2017-04-06T12:30:17Z

One option would be to take over and make another PR with changes on the top of the ones made here (pull this branch, make changes, make another PR).
Another option would be to push to this branch directly, but that would require push rights from @teonbrooks .

teonbrooks · 2017-04-06T13:01:14Z

@rodrigohubner I can give you write privs to my branch if you would like, let me know.

for the LSLClient we are working on, I know that @aj-ptw has been working on making a compatible lsl interface with the npm module he's worked on.
also @alexandrebarachant has made an interface for the MUSE via LSL.
I think this should likely be our course of actions for realtime clients, relying on unified client and have devs work to their hardware compatible with those adopted standards (e.g. LSL, OSC, Fieldtrip client)

larsoner · 2017-04-06T13:02:43Z

One option would be to take over and make another PR with changes on the top of the ones made here (pull this branch, make changes, make another PR).
Another option would be to push to this branch directly, but that would require push rights from @teonbrooks .

Keep in mind that PRs can be made from any branch to any other branch -- we juts usually make them from user/branchname into mne-tools/master. But you can just as easily make a PR from rodrigohubner/rtclient-fixes into teonbrooks/rtclient. This doesn't require push access or taking over.

teonbrooks · 2017-04-06T13:07:37Z

Keep in mind that PRs can be made from any branch to any other branch

yeah, I agree, PR to my branch would work, and would be less hassle.

jasmainak · 2017-04-06T18:51:16Z

@teonbrooks how close are we here?

I am going to add a TODO at the top of the PR so we don't lose focus

jasmainak · 2017-04-06T19:01:35Z

mne/realtime/base_client.py

+
+from mne.io.meas import create_info
+
+def _buffer_recv_worker(client):


it's the same for all clients, no? can we put it in one file and import it?

jasmainak · 2017-04-06T19:02:42Z

mne/realtime/base_client.py

+        pass
+
+    def _enter_extra():
+        """For system-specific loading and initializing during the enter


system-specific -> client-specific

jasmainak · 2017-04-06T19:03:34Z

mne/realtime/lsl_client.py

+        """
+        inlet = pylsl.StreamInlet(self.client)
+
+        ## add tmin and tmax to this logic


still todo!

jasmainak · 2017-04-06T19:04:03Z

mne/realtime/lsl_client.py

+        ## add tmin and tmax to this logic
+
+        while True:
+            samples, timestamps = inlet.pull_chunk(max_samples=self.buffer_size)


samples, times to conform to MNE nomenclature?

jasmainak · 2017-04-06T19:04:37Z

mne/realtime/lsl_client.py

+
+        return self
+
+    def create_info(self):


Missing docstring

jasmainak · 2017-04-06T19:04:47Z

mne/realtime/lsl_client.py

+        self.buffer_size = buffer_size
+        self.verbose = verbose
+
+    def connect(self):


missing docstring

jasmainak · 2017-04-06T19:05:54Z

@teonbrooks let's try to finish it while the momentum is still strong :)

andrewjaykeller · 2017-04-06T19:14:05Z

yes @teonbrooks ?

teonbrooks · 2017-04-07T22:45:26Z

hey guys, agreed. I was flying yesterday back to SF so I didn't see any of this. I will address these comments this weekend :) also @aj-ptw, is it straightforward to establish a LSL stream with the npm package? maybe we can chat tomorrow to make discuss what it would take in a simple example

codyworld · 2017-04-20T16:36:33Z

Hi all, I have questions. I am able to stream EEG data from OpenBCI Cyton V3 hardware into MATLAB environment using BCILAB via LSL and I would like to polish the incoming data in real-time prior component analysis. I know BCILAB has the capability to perform data cleaning using filters like Clean_rawdata, PREP Pipeline, ASR, Infomax etc. Does anyone have any idea, how to perform those process online?

jasmainak · 2017-04-20T21:26:33Z

I would like to polish the incoming data in real-time prior component analysis. I know BCILAB has the capability to perform data cleaning using filters like Clean_rawdata, PREP Pipeline, ASR, Infomax etc. Does anyone have any idea, how to perform those process online?

You can probably use autoreject with the fit and transform API. The fit would be run offline and the transform would be run online. Depends how bad your data is. But if you're working with Matlab, I'm not sure how you could make it work. Maybe you can try calling Python from Matlab like here

larsoner · 2017-08-11T01:46:10Z

@teonbrooks @jasmainak what is the plan here? Any time to work on it?

jasmainak · 2017-08-11T09:39:24Z

I'll be seeing @teonbrooks end of this month for the CRN coding sprint. @teonbrooks do you think we could earmark a couple of days after the sprint to work on this?

teonbrooks · 2017-08-17T03:28:37Z

yeah, I think this can be something we can knock out while @jasmainak is in town :)

larsoner · 2018-05-16T17:22:50Z

@teonbrooks this is about a year old, still planning to work on this or should we close?

teonbrooks · 2018-05-16T21:06:12Z

let's close this for now. maybe when I get a weekend free...

jasmainak reviewed Mar 29, 2017

View reviewed changes

teonbrooks force-pushed the rtclient branch 2 times, most recently from bb3d809 to cd0e135 Compare March 29, 2017 14:40

rodrigohubner reviewed Apr 6, 2017

View reviewed changes

jasmainak reviewed Apr 6, 2017

View reviewed changes

mne/realtime/lsl_client.py

"""

inlet = pylsl.StreamInlet(self.client)

## add tmin and tmax to this logic

Copy link
Copy Markdown

Member

jasmainak Apr 6, 2017

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

still todo!

jasmainak reviewed Apr 6, 2017

View reviewed changes

mne/realtime/lsl_client.py

return self

def create_info(self):

Copy link
Copy Markdown

Member

jasmainak Apr 6, 2017

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Missing docstring

jasmainak reviewed Apr 6, 2017

View reviewed changes

mne/realtime/lsl_client.py

self.buffer_size = buffer_size

self.verbose = verbose

def connect(self):

Copy link
Copy Markdown

Member

jasmainak Apr 6, 2017

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

missing docstring

teonbrooks added the mne sprint label Apr 7, 2017

teonbrooks added 2 commits September 4, 2017 10:50

working on the realtime client

254e0a1

new base class for rtclient

6c8637a

teonbrooks added 5 commits September 4, 2017 10:50

updates to the base client

252d77d

add import

6758621

first pass at lsl in mne

dc8e84c

UPDATE lsl client, base_client, FIX epochs

eb1bff6

fix

438f432

teonbrooks force-pushed the rtclient branch from 6b65fbb to 438f432 Compare September 5, 2017 07:07

teonbrooks closed this May 16, 2018

teonbrooks mentioned this pull request Apr 4, 2019

[MRG] LSLClient #6120

Merged


		from mne.io.meas import create_info

		def _buffer_recv_worker(client):

Uh oh!

Conversation

teonbrooks commented Mar 28, 2017 • edited by jasmainak Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

TODOs

Uh oh!

jasmainak commented Mar 29, 2017

Uh oh!

Choose a reason for hiding this comment

Uh oh!

codecov-io commented Mar 29, 2017

Codecov Report

Uh oh!

rodrigohubner commented Apr 2, 2017

Uh oh!

jasmainak commented Apr 2, 2017

Uh oh!

rodrigohubner left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

jaeilepp commented Apr 6, 2017

Uh oh!

teonbrooks commented Apr 6, 2017

Uh oh!

larsoner commented Apr 6, 2017

Uh oh!

teonbrooks commented Apr 6, 2017

Uh oh!

jasmainak commented Apr 6, 2017

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

jasmainak commented Apr 6, 2017

Uh oh!

andrewjaykeller commented Apr 6, 2017

Uh oh!

teonbrooks commented Apr 7, 2017

Uh oh!

codyworld commented Apr 20, 2017

Uh oh!

jasmainak commented Apr 20, 2017

Uh oh!

larsoner commented Aug 11, 2017

Uh oh!

jasmainak commented Aug 11, 2017

Uh oh!

teonbrooks commented Aug 17, 2017

Uh oh!

larsoner commented May 16, 2018

Uh oh!

teonbrooks commented May 16, 2018

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

9 participants

teonbrooks commented Mar 28, 2017 •

edited by jasmainak

Loading