ABOUT
Kelly Norton is a software engineer living in Surf City, North Carolina. He
holds degrees from Georgia Tech's School of Electrical and Computer Engineering
and from the MIT Media Lab where he studied under John Maeda. He has co-founded
a couple of companies. Most recently,
Fullstory, where he works as a Software
Engineer. He also co-founded Connexxia LLC in the early 2000's to help
universities effectively recruit high school seniors through online social
engagement. He spent six years at Google, working on Google Web Toolkit, Speed
Tracer, Google Chrome and other cool stuff. He worked on
Etsy's product search and he also spent nine
years building MailChimp where he was a
Distinguished Software Engineer. He was a key member of the team that grew
Mailchimp to over $1B in revenue before it was acquired by
Intuit in 2021 for an astounding $12B.
Honestly, he really just likes to build things.
JOURNAL
I recently re-learned a thing about PostgreSQL that I had learned at least one
time previously. If you create a table with a foreign key constraint that
includes cascading deletes and/or updates, PostgreSQL and MySQL have differeing
views on whether the indexes to support those cascading operations should be
created automatically.
Recently someone shared a link to a Javascript quiz that asks various questions about Javascript's infamous quirks. One of the questions required that the reader have basic understanding of floating point rounding. This is not a problem unique to Javascript and I attempt to explain why.
How many bits are set to one in a bit string? It’s a problem that pops up in a number of areas: bitsets, cryptography, error correction, and most of all interview questions. This problem is also known by many names: hamming weight, population count, popcount, sideways sum just to name a few. I recently revisited this problem in the context of writing a simple bitset and found myself falling down a rabbit hole when I tried to answer the question: how is std::popcount implemented in C++20?
A thing I learned recently is that constexpr and consteval, in modern C++, are more powerful than I orinally thought. While I never write C++ professionally these days, I recently ported a sudoku solver over to modern C++ and eventually realized that I could compute one of the static data tables at compile time using plain ole C++ code. This is a post about that.