dybr

Another day.

I'm still on PD. But everything was quiet until about 19h our time (10AM in California). When the latency started getting up-up-up. But this later.

I thought I wrote that 2-liner, to get the data from JSON together with JSON. Turned out, no way. Some tests failed, and see how. When you refer to a field in JSON, you specify a path, like (__ \ "customer"\"name") etc. What I need is the whole json. I wrote (__). It compiles, and some tests pass. But then, the reader sees "an empty path" - and crashes. I told typesafe people, years ago, that they need category theory specialists in their library. They were clueless. They were saying they needed a compiler guy (I'm not one). So, they have a bunch of BS inside their play library, including a pretty incompetent approach to monadicity. Like in HBASE, same thing. When (speaking in SQL), select * from users where age > 200; will crash (since there are no such users (yet).

Oh, whatever. Posted a question on stackoverflow (I know, chatgpt will be clueless), and had a nice discussion with another clueless guy, but eventually came to a conclusion: I'll handle it manually. It's not hard. Just takes much more than 2 lines. Maybe 20 or 30.

Then started looking through email, because Davis said that while on PD, one has to go check the emails from customer support and from users.

We had a 1-1, I told them that I a) have one PR from last week, closing one case, and another from this week, closing the case I opened regarding statistics, and plan to maybe take a look at this case that I'm supposed to do this sprint.

After the meeting, discovered a comment from Davis, on our team's channel in slack, telling me what I should do and what I should not do. I should not do what's in my sprint, because I'm on PD, but instead I should look for cases that belong to support, for example, he recently created a case, sent a pr, it did not work, so he rolled it back. Now, he says, I should take this case and implement proper caching. (I know, I know, we use guava cache, which comes from Google, from the guava cache that I wrote eons... 15 years ago - then it was rewritten by some guy called Fry.)

Sure can do. Except that he could have asked me. With respect. Maybe casually. Like, "Kirsha, could you please take a look at this case, I need help", or something like that. I'd have done it with pleasure.

Oh, and who is Davis? Is he my manager? He's nobody's manager. Just an engineer.
So I just thanked him, saying that I appreciate his attention to my work.

But sorry, now I'm just busy on alerts. Coming with an increasing frequency. The same latency thing. That "should not be logged", because we have too many logs. Yes, I did deploy to staging the small fix that disambiguates between two operations - but I did not have time to even look into staging's logs. Maybe they are ok. Then I can deploy to prod (thank you Safa and Swathi for approving fast). Then I can at least partially disambiguate, what kind of requests are causing troubles. The same as Friday. The same time, around 10AM PST. When Bay Area engineers and students are driving to work and turn on our app to listen.

BTW, why not log the location, for instance? Weird. We could.

Anyway, this was a good reason to relax for the day. Have to get up early tomorrow. And it's 20:41 already here.

Bonne nuit ! A demain !