I have a more mixed approach.

Decisions or attempts, i try to document them as comments on the ticket. It feels a bit like a conversation with my future self 😄

Things that pop up in my head that I don’t want to forget, but don’t have time to write out, I put in a new tab on nodepad++

Archi things go as markdown in de source code

Investigations and reports in a wiki

All these have all the pros of an engineering handbook out of the box (date, issuer, …)

I only use paper for the stuff I don’t want a trail off and just want to throw away afterwards.

In c they do indeed just mean shorter and longer int as the size of the int is defined by the compiler and target and originally represented the hardware.

There are types like int32_t or int_least16_t.

In later .net version for loops can be slower than linq queries. If the compiler knows the iteration does not cause side effects, it often already does not allocate and can even vectorize some loops, so .Sum(…) might be faster then a manual for loop.

Optimizing without benchmarking is often wasted effort.

I think they especially make sense for records so you can easily create shallow modifyable value semantic types with little boilerplate code.

I do hate that all these features are yet more keywords and weird syntax. It’s becoming C##.

Like the records are just sytesized methods on classes or structs. We already have this in the language with source generation. If something there was missing to create records, it would IMO make more sense to improve the source generation so the community also has access to the ecmxtensions. And the record could be split up in it’s separate features (copy constructor, defaul ToString imp, equatable) so we could choose to use part of it.

I bought a TV running VIDAA. You can use it without wifi and disable everything and let it just start on an input.

I don’t know of it’s EU only.