I also worked with japanese, including on site in Tokyo and quickly learned that asking "did you understand it?" is useless. I always had to keep in mind to ask "what did you understand?".
Cool project, but seems to be abandoned. At one point I was a subscriber to their premium version, but then started getting spam to the (unique) email address I used for the subscription. I emailed them to warn that their account database might be compromised but never heard back from them (this was back in '22).
Also, back then, their map tiles loading had a very high failure rate when loading, so I wrote a custom caching proxy to make it tolerable (which had built-in retry and also cached any successful response for a very long time).
No, he was an old man who cared for his wife with dementia until his death, an experience which changed him. And thus he has chosen to go on his own accord.
Spectre was shown to be exploitable from Javascipt: https://www.zdnet.com/article/google-this-spectre-proof-of-c... - making the bet that this won't be shown the same is not a safe wager I would say :) (especially that Javascript also includes stuff like WebAssembly).
Another reason for NOT self hosting: I want the passwords for my family to be available in the event something happens to me. The probability of Bitwarden being more resilient (at least mid-term) is much higher than any self hosting solution I would come up with.
I lead a migration very similar to this. We had taken over a system that was handed to a business person with all the technical team leaving because of a dispute. We rewrote the system and migrated all the user accounts over to the new system in a couple of months, after which we ran it for a couple more years, until the company got sold.
I would say the most important learnings were:
- there is lot of "extra stuff" around the product that is somewhat independent of "how complicated the product is". Even a simple CRUD app needs source control, a test suite, (ideally) some automated system for source quality check (like linters, analyzers, formatters, etc), a system for deploying it to a dev / staging / production environment, etc.
- production also needs monitoring (both "is it working" and "is it working within the expected parameters" - for example is it fast enough). Ideally there would also be some alerting around this monitoring so that you don't have to wait for users to complain to find out that something is not working.
- there is a saying of "use boring technologies" (https://boringtechnology.club/), which I 100% subscribe to. That will ensure that there are lots of examples for each aspect of the product you're trying to implement (for example authentication and authorization, creating an admin dashboard, etc).
- In addition I would say "use some managed platform to offload lots of these worries". Yes, it will seem weird to look at the bill at the end of each month and say "why are we paying $Xk each month for Heroku when I can rent a server from Hetzner for less than $100?" - but managing that Hetzner server (and probably more than one server, to make sure that a single hardware failure doesn't take down your entire product), ensuring backups are working, etc would cost more. Optimizing between "buy" vs "build" is a delicate balance.
In the end I think programmers who like to start new projects are a rare bread. I'd be happy to chat about this more (I'm also in Europe). Feel free to reach out to me at cratt[at]grey-panther.net.