Microsoft: let's develop a solution that helps *nix users use the best of two worlds! Let's create WSL! And then WSL2!

Also Microsoft: oh, let's randomly pick an IP for the WSL host so that it can mess with VPNs!

https://github.com/microsoft/WSL/...

Once again, the next build after hitting Ctrl + C while building a project surprised me with `java.lang.IllegalArgumentException: Malformed \uxxxx encoding`.

In which file? Is it so difficult to provide users with helpful information? Your `-X` option still does not show anything.

(Not to mention it must not corrupt ~/.m2/repository on SIGINT).

Timezones. So, general rules are:

1. If you don't store timezone, always use and assume UTC. Databases, backends, whatever you use, all time must be kept be in UTC.
2. If you store timezones, ensure you store them everywhere and don't drop them anywhere.
3. It's always better to ignore backend server time in favor of database's `now().` Having a single source of truth makes time consistent (if it's the same database, obviously). If you combine backend time and database time, you likely get a violation of causality.

I've just spent a couple of hours investigating "weird random one-hour time drifts on updates." Guys violated all three rules above:
- they didn't store the timezone;
- their servers had inconsistent timezones. Java was in +XX., while the server itself in UTC. On one host, they forgot to put JVM in the same timezone;
- they dropped the timezone because they thought it was the same everywhere, so there was no point in serializing it.

First of all, I hate crammers so much. These people kill the industry without even understanding it. They turned interviews into exams, missed the point of hiring, and saw no distinction between knowledge and information all the time. They don't understand that if you can google an answer in five seconds, it's not knowledge. It's information.

They don't understand that questions like 'what will Python do if you delete an item from a dict while iterating over it' are complete nonsense. They don't understand that it's not 'dig deep'; it's just a bad practice that leads to errors, thus must be avoided. The fact of remembering 'RuntimeError: dictionary changed size during iteration' means that you haven't been avoiding it enough.

One more example. Which signature is correct?
- ApplicationListener<ContextRefreshedEvent>
- ApplicationListener<ContextRefreshEvent>
- ApplicationListener<RefreshedEvent>
- ApplicationListener<RefreshEvent>

Second. What's the point of forcing you to write compilable code in google docs? Do they really expect that one could possibly remember 'import org.springframework.beans.factory.annotation.Autowired;'? Seriously?

Third. Why do they expect me to know Spark, Java, J2EE, Spring Boot, Python, Kafka, Postgres, React/Redux, TypeScript, and work for miserable 70K EUR?

What's wrong with the European IT job market? Are they fucking nuts?

I have to reimplement a couple of complicated OOXML parsers (docx, pptx, vsdx, etc). Actually, I’ve implemented them in Python ~5 years ago but now I have to improve them and add support for nested/embedded formats and some other stuff. As you could expect, none of the OOXML validators are valid themselves, so it's better to have an MS product installed locally, just to get reassured that everything works fine and the parser produces the format that's recognizable by M$.

So I’ve bought a key on eBay (yep, I’m not paying full price for this shit: release valid validators first, bitches; don't make me buy things I don't need). The key is valid, everything is fine. But no, you just cannot have a link to download this fucking installer, no-no-no-no. We won't give you a link until after you enter a key. FEEL DEPENDENT. OBEY.

But I digress. Here's their MANUAL about DOWNLOADING the INSTALLER:
https://support.office.com/en-us/...

So, what's wrong with it? Oh, just a minor misunderstanding. They always give you a link to download an exe-installer. Even if you use Safari.

Why everything is so fucked?

I really hate people that don't document well their 'inventions'.

Let's take for example kotlinx.serialization library. It's a self-obvious fact that there are at least two things that will make anyone suffer: generics and polymorphism. So, they must be on the front page. Solved. With a recipe. Ready to use.

And what do we have? One mention somewhere in docs and an amazing test that 'explains' 'everything':

https://github.com/Kotlin/...

'What we got here is a failure to communicate.' (c)