Hauleth

GitHub - appunite/mockery: Simple mocking library for asynchronous testing in Elixir. Simple mocking library for asynchronous testing in Elixir. - appunite/mockery

It doesn't seem to be active much, but github.com/appunite/moc...

Hauleth's blog - Jeopardy! world Blog about BEAM, Rust, fantasy & stuff

Something that come to me some time ago, but only now I have found time to write it down properly

hauleth.dev/post/jeopard...

#AI

Yeah, people do not understand what `NULL` in DB means exactly.

On the other hand I do not see the point of doing review of code that is incorrect. The same with manual testing which is more expensive than CI time. So I prefer to have full CI run before I start with manual testing or deep code review.

I mean, that is the point, isn't it? We need to check if change in one component didn't affected seemingly unrelated another component…

There is GHA step for that, but the problem is that it constantly recompiles my project over and over again and I do not know how to tell it to not do so, as it takes *ages* to compile.

I tried to setup pipeline that compiles some Rust code and cache it between action. 3h of GHA credits and it still doesn't work. I hate debugging that.

Typical GHA debugging session.

You mean calls to functions that aren't there yet?

Actually - plan is for both ;)

Image showing river with sign next to it with name "Bug" Over the image there are 3 lines of text, from top to bottom: It's not a bug It's not a feature It's a river in Poland

So something is hatching.

It is a tease of incoming project I am working on with a friend.

Yes, this is `psql` session, but it is *not* Postgres.

#ElixirLang

Just disclaimer - it is PoC, I never used that in any project.

If the FSF calls your program malware, is that a sign that you've made it?

Just add in the title that it is remote within US

Another useful feature unfortunately got closed, for reasonable reasons, but still, this is slightly painful to see that it will not be there, as it would help with having some fancier socket features available.

BIF for more detailed process information · Issue #9265 · erlang/otp Is your feature request related to a problem? Please describe. In systemd integration library it would be useful to have access to process effective user ID and process effective group ID. Unfortun...

Place where it would be useful is my systemd library where I need to send extra information about process when these differ. But with plain Erlang I cannot get that info, that mean that I need to use external NIF (wich I also have written). There is issue to add that functionality

Oh, another useful thing - libffi-like tooling built in. So it would not require me to write whole NIF and require user to compile (or go through pain of cross-compilation) some C code to access functions like `getuid` or `geteuid` etc.

Unfortunately - no. This is related to Elixir syntax, not `with` construct. You can do something like:

with(
a <- foo(),
b <- bar()
) do
…
end

Or

with \
a <- foo(),
b <- bar()
) do
…
end

But AFAIK both these forms will be changed by formatter.

A lot of others pointed nice things so I will add other things

I want to have better alternative to ExCoveralls, ideally built in. Same for ExUnit reporting - built in JUnit XML output or with just compilation data - nice machine readable format that could be ingested by non-Elixir tooling

Is there anything I can do to check these errors? Or maybe you need some help with that. I am open to consulting opportunities, so I can take a look.

Has anyone achieved to make @codecov.bsky.social Test Reporting work with #ElixirLang? I am using junit_reporter lib and it seems that CodeCov is having some problems with that file. Unfortunately they just say that they reported some errors to their Sentry, but do not tell me what are these errors.

I use Nix on macOS and I am supper happy with result.

I have another potential co-developer, assuming it will be in Elixir or Go. I am not sure about Go, but in this case it probably would be a good choice, despite my dislike of Go semantics and idioms.

Always - no, but it is nice to have, as for example I am using vim-projectionist that can create file scaffold when I am creating new file. With proper correlation between modules and filenames I can have automatic generation of module and test base as soon as I open the file. This is handy

I have similar plan, for local-dev-only OpenTelemetry ingester and dashboard that would be single binary without any extra dependencies. I was thinking about using @duckdb.org as a data store, so user can have some persistence, but by default just use in-memory.

Nothing git.sr.ht/~hauleth/lan...

Yeah, but when I want to migrate I always forgot about all that and I am frustrated all over again.

I do not have issues with MyApp vs MyAppWeb, but I would prefer to have `MyAppWeb.Foo.Comtroller` that uses `MyAppWeb.Foo.JSON` than some magical and "non standard" naming scheme in Phoenix.