Richard Startin

What are the olive green and opaque grains in ppl? I had to describe a thin section similar to this in an exam a while back, I could make out the clinopyroxene and plagioclase, but guessed the black was magnetite and the olive green might be alkaline clinopyroxene.

Haha. Well if you’re ever in London or Cardiff I’d sit through a match with you, though as a pre-IPO alumnus you might be disappointed not to be in a box.

I’m leaving Datadog at the end of next month and plan to spend some time on personal projects, I’ll probably look at finding another job next September.

17-year-long Scottish study of 82,000+ adults finds commuting by bike reduces your risk of:
• early death
• hospitalisation for any reason
• cardiovascular disease
• cancer mortality
• treatment for mental health problems.

Yeah, but those LTNs and bike lanes are *really* annoying.

Real professionals do not plug random devices into their computers. Real professionals do "curl -sL totally-trustworthy-site.com/install.sh | sudo bash" instead.

I have some notion you’re from the West Midlands, so should have an insurmountable bullshit detection mechanism?

Citation · Issue #1 · Josehokec/ACER4CER I haven’t read your paper but was disappointed to find that it doesn’t cite my blog post on the design of the RangeBitmap data structure your evaluation artifact depend on, which I also implemented...

It was really nice to see my work used to reduce latency of complex event recognition queries by an order of magnitude in a paper in ACM, but I was disappointed to find I hadn’t been cited, hopefully this will be fixed.
github.com/Josehokec/AC...

TIL when you use "from module import function" python still loads the entire module. Saw this in my memory flame graph. I moved some functions with heavy deps to a separate file and it deleted a whole "flame" of memory allocations

I am indeed based in the UK. I don’t think I know the kotlin standard library well enough to make serious proposals about changing it, though the job description does look very exciting.

Sharing benchmark results is definitely an unsolved problem; “reproducible benchmarks as a service” does not exist either and should because people struggle to provision adequate benchmark environments. I just wonder what the TAM is for something like that though, small enough to preclude existence?

I don’t tinker much these days and haven’t solved a hard algorithmic problem in a long time. I see people working on databases (which surface lots of interesting problems) but honestly think *most* data sets belong in Postgres.

My personal technology consumption comes in two guises: I use apps to consume information or communicate, and I use python for data analysis on a regular basis. I generate the bulk of my python code these days, before cosmetic refactoring.

I was going to say too bad the locations are so restrictive, but my kids are half Serbian but don’t speak the language, we were considering moving somewhere in the balkans for a while to help them learn, which would have meant a sabbatical for me. Maybe I should move and put in an application? 🤔

Kotlin Libraries Senior Software Developer (Kotlin) - Kotlin Libraries - Careers

Kotlin libraries team is hiring: www.jetbrains.com/careers/jobs...

If you are eager to write something you are going to regret (if you are lucky, also be proud of) in 10 years, want to tinker with Kotlin and not afraid of dragons (also: performance, concurrency), drop me a line!

Which foundational problems in technology need to be solved? I know lots of people want to solve certain foundational problems, but which ones need solving?

Why it had to go
I don’t know, the logs didn’t say
Something went wrong
Now I long for yesterday

I can't tell you how many hours I spent wandering around some local woodland on this day, but light hitting this branch saved me going nuts

What sort of mathematical techniques are you using for the modelling? PDEs? (Presumably non-stationary) stochastic processes?

My issue was with the tracing paper rather than stereographic projection. I would have been more engaged if this had been computer based, and we’d needed to think about how to actually compute a best fit curve for a fold rather than do it by eye, but that’s just the sort of teenager I was.

I was generally interested in geology, but I found course content like using stereonets (on tracing paper!!!) for structural geology incredibly frustrating when I really just wanted to find out how much calculus/algebra/whatever I could do at that time.

I had the impression (as a stupid 18 year old) that if I really wanted to be a geophysicist one day, I could do that with a maths or physics degree (I didn’t), but the reverse wasn’t true and I felt it would be treated with suspicion in careers which required generic quantitative skills.

I started a geophysics degree about 20 years ago but switched to maths. I felt the degree was too specific. Less geology than a geology degree, a much lower standard of physics than a physics degree, not as quantitative as I’d been led to believe (this was important to me).

I’m curious what kind of mathematical techniques you use for earthquake forecasting. My first degree was in maths but I’m taking a part time degree in geology as a kind of hobby, something in this area might make an interesting project.

It only tracks that for similar runs, so if you run a particular route frequently it’ll give it to you under matched runs, e.g.

This would be really helpful especially when you’re trying to vary one of the parameters by more than an order of magnitude to plot line charts

So what I was missing was what the start stack trace actually is. It makes sense now.

I find the approach of using coins for scale a bit strange (as a geology outsider) compared to say, a 6’’/15cm ruler, because I have no idea what the dimensions of Norwegian coins are.

I’m probably missing something obvious but why would anyone ever want the start stack trace?

I got incredibly excited about the vector API in 2017, long before I realised I would probably retire from the industry before being able to use it in production.

I recall having a conversation with a former colleague in a pub about how backticks was the right way to do it and $(…) was somehow implementation specific. Maybe too many pints had been drunk by that point.