v0.0.3

[scpwais]

In this revision:

• Major clean-up / re-structure of python protobag
• Python and C++ examples in the examples subdir
• A demo python notebook

@rreusser @Erkaman @dustindorroh @cczarnocki would love your feedback on the python notebook demo. For now it's easiest to just read the version stored on Github (versus try to run it). Here's a permalink to the latest revision at time of writing: https://github.com/StandardCyborg/protobag/blob/aed9ea01d91eaecc838972a0a85bfc4d79b0d6d3/examples/protobag-to-parquet/protobag-demo-full.ipynb

[scpwais]

@Erkaman @rreusser @dustindorroh @cczarnocki

The python notebook now runs on Colab! Since the repo isn't open source, you have to do a few things to get it working:

• Download the notebook (here's a perma link: https://github.com/StandardCyborg/protobag/blob/60a0e0a2b5e82c2520579994fea27be21ad3b26b/examples/protobag-to-parquet/protobag-demo-full.ipynb )
• Download the python wheel and protobuf definitions I have saved for you here: https://drive.google.com/drive/folders/1zcX9eRU4R5GPZdJfCHoUa6WZPULxWc9q?usp=sharing
• Go to Colab and go to File > Open Notebook https://colab.research.google.com/ . Choose Upload and upload the notebook file protobag-demo-full.ipynb that you downloaded.
• Make sure you Colab notebook is connected to a VM (in the upper right hand side you should see a Disk/RAM icon, or a button that says Connect [click that button!]). Now upload the python wheel and MyMessages pb file to the Colab VM: in the left hand side of the notebook there's a folder icon, and if you click there it'll show you the local filesystem and there's an upload button you can use.

Now you can go to Runtime > Run All to run all cells, or run them one by one, or whatever you want.

[scpwais]

Added python support for using the TimeSync utils. Added a demo of using it to the jupyter notebook demo: https://render.githubusercontent.com/view/ipynb?commit=293ae8ee82510f3d207d22317997d5f432827e14&enc_url=68747470733a2f2f7261772e67697468756275736572636f6e74656e742e636f6d2f5374616e646172644379626f72672f70726f746f6261672f323933616538656538323531306633643230376432323331373939376435663433323832376531342f6578616d706c65732f70726f746f6261672d746f2d706172717565742f70726f746f6261672d64656d6f2d66756c6c2e6970796e623f746f6b656e3d414f334c5647374b41454d3246494a4d4a344e4242484b37454e513649&nwo=StandardCyborg%2Fprotobag&path=examples%2Fprotobag-to-parquet%2Fprotobag-demo-full.ipynb&repository_id=258341093&repository_type=Repository#Synchronizing-Time-Series-Data

(edit: above link might be broken, here's a better perma-link https://github.com/StandardCyborg/protobag/blob/293ae8ee82510f3d207d22317997d5f432827e14/examples/protobag-to-parquet/protobag-demo-full.ipynb )

that link might not send you to the right section. search for "Synchronizing Time-Series Data"

[Erkaman]

"The user provides as parameters a 'max_slop' (a time duration) and a 'max_queue_size' (a number of messages to buffer"

buffer? what does this buffer do? its the first time its even mentioned.

[Erkaman]

"If topics have messages recorded at somewhat different times (e.g. two sensors both record at 5Hz, but one records 50ms before the other), then MaxSlopTimeSync will find the nearest matches."

ok. im interested in how it actually does this. how finds the nearest matches. please write down a short summary of the algorithm used. otherwise, the thing feels like a black-box to me, which i dont like,

[Erkaman]

I can definitely see the utility of using MaxSlopTimeSync for syncing up depth and rgba data captured from the iphone sensors. 👍👍👍

[Erkaman]

spec.set_max_slop(seconds=2, nanos=0)

what does the nanos=0 argument do?

[Erkaman]

@scpwais left all my feedback above.

[scpwais]

thanks @Erkaman ! Linked in what you read is the MaxSlopTimeSync docs; here's that link again: MaxSlopTimeSync I think that will answer most of your questions. You might also consult the unit tests for MaxSlopTimeSync and for IterProducts. I'll update the notebook a bit to make these more conspicuous.

[scpwais]

"what does the nanos=0 argument do?" -- nanos is nanoseconds