jorin @jorin@mas.to

What's happening to https://softwareengineeringdaily.com/ ? Too many strange episodes popping up today. Had to unsub. Was really confused but after some research it seems Jeff is not doing well right now. Hope he gets better soon 🙏
https://news.ycombinator.com/item?id=28504276

It's not about what's possible with a language/platform but what's the default

UI test assertion fails. Event triggered re-render but nothing changed. Immutable data structures allow basically using React.PureComponent by default.

Wrong field matching data structures. 2 fields called "type", different meaning. Wouldn't happen with qualified keywords.

Already ran into 2 bugs today that would not exist with #clojure stack:

> People talk about "validating" their business ideas by doing market testing, but "validating" has a bias towards a positive outcome. In reality the outcome is usually negative. So don't validate your startup ideas, invalidate them.
@mccrmx@twitter.com
https://mccormick.cx/news/entries/how-i-invalidated-my-best-startup-idea

Security and privacy are so painful to wrap my engineering brain around.
These two areas are all about "good enough" from the very start. Nothing will be 100% secure or private.
With the rest of software we can at least pretend to have found the perfect solution.

A day goes by. Waiting for CI.

One of the worst things about Confluence is that I have to decide where to put the page in the hierarchy before I wrote even one word.
I much prefer organic structures. Both for knowledge and organizations.

Super excited about OpenTelemetry. And also: HTTP is pretty awesome. Love that the protocol's extensible design still holds up today. HTTP/2 and the Trace Context seem to work great for modern service architectures.
https://doordash.engineering/2021/06/17/leveraging-opentelemetry-for-custom-context-propagation/

Saw a great thread on this today:
https://twitter.com/Julian/status/1404545905483681796

When building software I tend to jump to trying to find a solution to the problem. Instead I should ask the more crucial questions first:
How can I be confident the solution works as intended?
How can I have a productive dev workflow while solving and evolving the solution?

Of course there are exceptions. You can get rich by having a patent for some fancy crypto algorithm. Not my cup of tea though. Feels morally similar to profiting from a patent for a vaccine.

A project or company is a system. Code is only a momentarily artifact the system produced. If that artifact is still all the value you have, the system is dead.

If you are scared someone stole your business because they saw your source code a few years ago, then I argue you don't know what is the value you deliver. Code is one of the least important bits :)
https://intuitiveexplanations.com/tech/replit/

We rarely do stuff like restarting a web server on a new port once a new config event came into the DB. Maybe we should do more of that.

Seems like a really solid platform to build on. I would be curious to find out more about the tradeoffs it takes.
It seems to be mainly designed for client-side applications. What tradeoffs does it take compared to for example the JVM or Go's design?