Clojurians Log v2
clojuredesign-podcast
How does exploring via the REPL help you speed up application development? What are you learning as you explore? How do you interact with large data sets? In our latest episode, we realize we're done exploring when all of the pieces fall into place. https://clojuredesign.club/episode/104-assembleify/ What is the most interesting thing you've explored via the REPL?
I'm really enjoying the new series and it's amazing that you have created over 100 episodes I love the names of the new episodes and am wondering if there will be an episode on testing... hopefully in the gospel choir style as I assume it will be called testify 😁 https://youtu.be/L8MAe9HbePE?si=l7aX81Hv69ln4Tii
Oh, that’s too good! Testify!
@jr0cket I love the topic idea and the title is perfect!
@jason.bullers That’s a fantastic example! Thanks for sharing. That reminds me… When Nate and I were working with esports data for a big league, we would run every game ever played through the system to look for regressions. Over the years, that’s was many GB of data and some really interesting edge cases!
The discussion around variance in the data reminded me of something from work years back. We were working on a system for hurricane forecasting. Because most storms come up from the south, Canadian meteorologists tend to just ingest and republish the track forecasts issued by our American friends, until the storm moves far enough north that it's in what's called the "response zone". At that point, a Canadian meteorologist will take the latest American track forecast and start issuing their own forecasts (and presumably the American forecasting ends). At the time, the way to ingest the American tracks was to read in and parse a text bulletin (I'm sure you can already guess where this is going). We had a bunch of samples, with bulletins starting with phrases like "Tropical storm six ..." or "Hurricane Chris ...". Pretty straightforward regexes. Then we heard a rumour about how the legacy system we're replacing choked once upon a time because there was an American bulletin with the non-standard phrase "Remnants of tropical storm six ..." We went searching around, and sure enough, we found it. The guy on the team doing the parser wisely decided to play it safe and found an archive of the last decade of bulletins. He fed it all through the parser, and discovered several "remnants of" bulletins, along with a few other anomalies. "Remnants of" is now our go-to phrase for covering your butt when parsing. Be as permissive as possible!
Such a great story!! And also highlights the enormous value of archived data. It can teach you so much.
Thank you for sharing!