a11r

The bulk of the tiger population is in the Sundarbans (marshlands at the bottom of the Ganga-Brahmaputra delta). Anecdotally, many of the Sundarban tigers are man-eaters and there is a tacit policy to not go after the ones that happen to kill humans. This keeps the poachers and encroachers away, but also makes any census nearly impossible to conduct. I would not put too much faith in these reported numbers.

I brute-forced a 32-bit random number generator in an internet worm, back in the day[1]. Brute force is underrated.

[1] http://www.icir.org/vern/papers/witty-imc05.pdf

Most tests are probabilistic and you have to think about the conditional probability given other indicators. Consider a urine culture for detecting UTIs. If you just get urine samples from a general population, some fraction will show positive (bacteria in the culture) falsely. It would be pointless to perform this test on the general population and to treat people that show positive . But if you perform this test conditioned on the fact that the patient is presenting other symptoms (burning sensation, fever etc.) then the same test very useful and the conditional false probability is low enough that it can be used to decide on treatment.

The same idea applies to most tests in medicine. That is why tests are mostly done conditionally.

Also, the population growth rate in India was significantly lower during the colonial period (1750 to 1950). This was primarily due to increased mortality from the famines created by the British. Over a 200 year period, the British colonialists performed the largest instance of genocide known to humanity.

A simple statistic: Before the British showed up, Indian GDP was 15 to 20% of world GDP. When they left its was 1.5 to 3% of world GDP. Something tells me the colonial experience might have something to do with it.

A great account of the building of the IMP: "where wizards stay up late".

I am assuming that you are talking about the thermal expansion of the tubes (ducts ?). A long time ago, the rails making up train tracks used to have gaps every few meters to accommodate thermal expansion (same problem). Modern rails in high speed train tracks are welded into one continuous bar that can be over a kilometer long. The trick is to stretch the rail just enough and then pin it down - kind of like a slightly stretched rubber band. That stretch can accommodate thermal expansion.

gRPC uses the following features from HTTP/2: Framing (needed to multiplex different RPCs on the same TCP connection), Flow Control (both intra-stream and connection-wide), HPAC (for compressing metadata that tends to be common across requests), and GoAway (to gracefully drain backends).

Additional IP Rights Grant now added to the repo: https://github.com/grpc/grpc/blob/master/PATENTS

Benchmarks are under development: you can see the code in https://github.com/grpc/grpc/tree/master/test/core/fling and https://github.com/grpc/grpc/tree/master/test/cpp/qps .

You are correct in assuming that, technically, gRPC could be built on top of WebSockets. Building directly over HTTP/2 allows for a leaner protocol stack.

Protocol Buffers didn't really have an RPC engine - the services feature was limited to generating some interfaces. gRPC provides an actual RPC engine with required network functionality.