grpc

Background

gRPC is a description format and library for services built on top of Protobuf. I wrote an R wrapper to the core C library in 2017, but it is lacking several key features:

• Authentication - be able to provide / accept authentication tokens
• Security - use TLS to encrypt messages
• Streaming - reuse an existing connection for multiple requests

Let me know what you think.

Related work

http://github.com/nfultz/grpc is a basic implementation.

https://github.com/eddelbuettel/rprotobuf - Protobuf for R

https://github.com/grpc/grpc - underlying C library

Details

The student will add and extend the functionality in the grpc package from both R and the underlying C wrapper code.

There are two main code paths in the grpc system - grpc clients / senders and grpc servers / receivers. The student will add features to each.

Security - gRPC can use TLS to encrypt the messages sent. This needs to be configured when a new connection is created.

Authentication - grpc also supports creating and sending authentication tokens (such as Oauth2 tokens) as metadata attached to each message.

Streaming - support sending multiple messages over a single connection. Because sessions may be long-lived, the server may need to fork to handle multiple requests.

Expected impact

Allows R to interoperate with other gRPC projects. Apache Arrow is one, as are many Google ML products.

gRPC is also used pretty widely in industry as a language agnostic way to connect microservices.

Mentors

• EVALUATING MENTOR: Neal Fultz <[email protected]> is the author of R packages grpc and stackoverflow. He is the data science lead at UCLA Social Science Computing and the principal of NJNM consulting, and coorganizer of the Los Angeles R User Group from 2010-2014.

• SECOND MENTOR - Gergely Daroczi <[email protected]> is the author of eg pander, AWR, botor and logger packages. He is a Director of Data Ops at System1, and the organizer of the Budapest Users of R Network, a satRday in 2016 and the European R Users Meeting 2018.

Tests

Students, please do one or more of the following tests before contacting the mentors above.

• Easy: Write an Rcpp function which takes an integer, n, and an R function, callback. The Rcpp function should draw n samples from the normal distribution, and execute the call back function on each sample that is greater than zero.

• Medium: Install the grpc R package from source, or rebuild the docker container on a newer version of debian.

• Medium: Modify the grpc client timeout to take it's value from options(). See also https://github.com/nfultz/grpc/issues/28

• Hard: Build and check the grpc package on Windows. See also https://github.com/nfultz/grpc/issues/21

Solutions of tests

Students, please post a link to your test results here.

• EXAMPLE STUDENT 1 NAME, LINK TO GITHUB PROFILE, LINK TO TEST RESULTS.

K. Kahurani, Test results

Ashraf Ali, Test Results

Shehab Hosny, Test Results