--- a
+++ b/docs/notes.md
@@ -0,0 +1,40 @@
+## Cloudbrain data structure when publishing and subscribing:
+```
+[
+  {timestmap, channel_0, ..., channel_N},
+    ...
+  {timestmap, channel_0, ..., channel_N}
+]
+```
+
+## Cloudbrain style guide
+* 4 space indents
+* 80 characters per line max
+* package names should all be lower case
+* limit number of classes to 1 per package if possible
+* stick to underscore for module names, repo names, config files, etc. Try to avoid hyphens. Camel case is for class names.
+
+## TODO HTB
+[x] Deploy new notch filter
+[x] FrequencyBandTransformer module
+[x] Discuss FFT vs STFT w/ manual chunking with pierre (fft_test.py VS stft_test.py)
+[ ] better subsampling in RT server
+[x] publish / subscribe tutorials
+[x] bring back the timestamps in FrequencyBandTransformer module
+[x] band pass that makes sure to avoid the 60Hz freq
+[x] StdoutSink
+[x] PyPlotSink
+
+
+## TODO later
+[ ] Update or rename PipePublisher. Right now it's just a stdout publisher and it's unlikely 
+that we'll go down the named pipe route.
+[ ] websocket publisher and subscriber
+[ ] rest source (for label input)
+[ ] lsl publisher and subscribers
+[x] 2 filter modules: bandstop and highpass
+[ ] Rename FrequencyBandModule to FFTModule?
+[x] Make a sink module that is plotting the chart
+[ ] Add the simple threshold classifier
+
+