{"title":"uncomfy","subtitle":"uh... i have a website?","link":[{"@attributes":{"rel":"self","type":"application\/atom+xml","href":"https:\/\/uncomfyhalomacro.pl\/atom.xml"}},{"@attributes":{"rel":"alternate","type":"text\/html","href":"https:\/\/uncomfyhalomacro.pl"}}],"generator":"Zola","updated":"2026-04-07T05:06:31+00:00","id":"https:\/\/uncomfyhalomacro.pl\/atom.xml","entry":[{"title":"Part 7: So I am building a fullstack project","published":"2026-04-07T05:06:31+00:00","updated":"2026-04-07T05:06:31+00:00","author":{"name":"\n Soc Virnyl Estela\n "},"link":{"@attributes":{"rel":"alternate","type":"text\/html","href":"https:\/\/uncomfyhalomacro.pl\/blog\/202604130543\/"}},"id":"https:\/\/uncomfyhalomacro.pl\/blog\/202604130543\/","content":"

It's been a while, so I kind of forgot to write this series.<\/p>\n

Removed Deno Fresh\u00a7<\/a\n>\n<\/h1>\n

While developing the project, I've noticed a disadvantage of using Deno Fresh.<\/p>\n

The disadvantage is Deno Fresh is not well-maintained. The only developer<\/em> that\nwas supposed to maintain it resigned.<\/p>\n

Hence, it's not really a good idea for me to write something in Fresh. Sadly, it is what it is.<\/p>\n

Branching out for a new project\u00a7<\/a\n>\n<\/h1>\n

I've noticed some pattern I've been doing repetitively\u2014writing migrations.<\/p>\n

I was thinking of writing a project similar to alembic. I know SeaORM might\nalready have it but I was curious what the thought process be like in building\nsimilar to alembic and SeaORM.<\/p>\n

Recent changes\u00a7<\/a\n>\n<\/h1>\n

I recently fixed some issues on how I handle cookies in axum between backend and client. I've\nbeen using private cookies over signed or plain cookies.<\/p>\n

Another similar project\u00a7<\/a\n>\n<\/h1>\n

I also am refreshing my FastAPI knowledge in\nnews-portal<\/a> and\nmgmt<\/a>.\nmgmt<\/a> is the basis for my\nfullstack-fastapi-template<\/a>.\nI separated the projects, mgmt in GitHub and the template in Codeberg, cause\nmgmt is LLM assisted, specifically, the frontend is generated with LLM to\nexperiment with the backend responses. Not my cup of tea but well, it is what it\nis as recruiters are looking for AI experience.<\/p>\n

Ah what are you doing now?\u00a7<\/a\n>\n<\/h1>\n

I don't know much. Currently, I've been looking for work, and it seems it's not\nreally working out for me despite my experience in Rust and Python. The job\nmarket is pretty saturated with LLMs as part of the development process right\nnow, and my degree and background might be a problem. I even got silently\nrejected by a local company in my city. But it is what it is. I'm confident that\nI will get a job but not around this month. Not even sure if May is a good month\ntoo.<\/p>\n"},{"title":"Part 6: So I am building a fullstack project","published":"2026-03-23T00:19:10+00:00","updated":"2026-03-23T00:19:10+00:00","author":{"name":"\n Soc Virnyl Estela\n "},"link":{"@attributes":{"rel":"alternate","type":"text\/html","href":"https:\/\/uncomfyhalomacro.pl\/blog\/202603081856\/"}},"id":"https:\/\/uncomfyhalomacro.pl\/blog\/202603081856\/","content":"

For the past few days of improving backend and experimenting axum with utoipa, I\ncame to a conclusion that you can offload core logic to regular functions\nfor reuse.<\/p>\n

\n

Here is a Rust pseudocode of what I mean<\/p>\n

pub<\/span> fn<\/span> login<\/span>(<\/span><\/span>\n\tState<\/span>(<\/span>state<\/span>)<\/span>:<\/span> State<\/span><<\/span>Arc<\/span><<\/span>AppState<\/span>><\/span>><\/span>,<\/span><\/span>\n\tJson<\/span>(<\/span>login_details<\/span>)<\/span>:<\/span> Json<\/span><<\/span>UserLoginRequest<\/span>><\/span><\/span>\n)<\/span> -><\/span> Result<\/span><<\/span>axum<\/span>::<\/span>response<\/span>::<\/span>Response<\/span>,<\/span> StatusCode<\/span>><\/span> {<\/span><\/span>\n\tlet<\/span> user<\/span> =<\/span> login_service<\/span>(<\/span>login_details<\/span>)<\/span>.<\/span>await<\/span>.<\/span>map_err<\/span>(<\/span>|<\/span>err<\/span>|<\/span> {<\/span><\/span>\n\t\ttracing<\/span>::<\/span>error!<\/span>(<\/span>?<\/span>err<\/span>)<\/span>;<\/span><\/span>\n\t\tStatusCode<\/span>::<\/span>INTERNAL_SERVER_ERROR<\/span><\/span>\n\t}<\/span>)<\/span>?<\/span>;<\/span><\/span>\n\tOk<\/span>(<\/span>Json<\/span>(<\/span>user<\/span>)<\/span>.<\/span>into_response<\/span>(<\/span>)<\/span>)<\/span><\/span>\n}<\/span><\/span><\/code><\/pre>\n
Currently, my code smell is actually passing the functions that process form data\nto a function that process json data when I was experimenting form based logins\nwith reverse-proxy and it hit me\u2014I should separate core business logic when handling\ndata. Oh well, I'll be on a huge refactor, mostly separating similar logic into a services\nmodule so I won't repeat myself.<\/p>\n<\/blockquote>\n
For example, if you have function in public<\/code> and admin<\/code> modules that handle\nlogins, instead of naming them both as login<\/code>, name them separately as\npublic_login<\/code> and admin_login<\/code>. Another workaround I have was hard-coding the\npaths.<\/p>\n
I actually combined both and I think I rather do the former since I think\nnesting should be reused here.<\/p>\n
However, the latter is still problematic as it is affected by the same bug if\nyou accidentally rename the functions to the same name so \/admin\/login<\/code> and\n\/api\/login<\/code> are merged as either \/admin\/login<\/code> or \/api\/login<\/code> whichever comes\nfirst when generating the openapi data.<\/p>\n
\n
So why talk about form data<\/strong>?<\/p>\n<\/blockquote>\n
I recently came across the usage of the formaction<\/code> attribute. It's basically href<\/code> but\nsends a form data request to the target path. Hence, I've been experimenting using caddy<\/code>\nto reverse-proxy both my frontend and backend to a common URL e.g. https:\/\/localhost:8888<\/code>.<\/p>\n
The downside is that, I won't be doing manual fetch<\/code> in the frontend client, meaning,\nthis approach is unable to use JSON as a request body for requests in the backend.<\/p>\n
Another downside is I am not able to use the routes that use JSON as the request body.<\/p>\n
However, I believe that the upsides far outweigh the downsides because we can always\ncreate a toggle for the frontend to use the JSON based routes. Another upside\nis the refactor, which means it's easy to swap Form<T><\/code> to JSON<T><\/code>.<\/p>\n
That's all for this project's update. I think I'll be taking a 1 day break from it. Too tired since\nyesterday.<\/p>\n"},{"title":"Part 5: So I am building a fullstack project","published":"2026-03-15T07:07:34+00:00","updated":"2026-03-15T07:07:34+00:00","author":{"name":"\n            Soc Virnyl Estela\n          "},"link":{"@attributes":{"rel":"alternate","type":"text\/html","href":"https:\/\/uncomfyhalomacro.pl\/blog\/202603134346\/"}},"id":"https:\/\/uncomfyhalomacro.pl\/blog\/202603134346\/","content":"
So I made some various experimentations with Fresh and Preact.<\/p>\n
Nothing really big but just some surface-level crap that I've been doing.<\/p>\n
What the Hell Even Are Signals?\u00a7<\/a\n>\n<\/h1>\n
I was trying to experiment with Tailwind CSS built into the\nfrontend template like for empty input or a status code that isn't\n201 or 200, specifically for the register and login modals.<\/p>\n
From my understanding, Signals looks similar to React's useState<\/code>.<\/p>\n
The usage is similar as well, but I am not sure what the differences are.<\/p>\n
Preact describes signals as an efficient way to update UI states.<\/p>\n
\n
In Preact, when a signal is passed down through a tree as props or context,\nwe're only passing around references to the signal<\/em>.<\/p>\n<\/blockquote>\n
This means that signals are efficient because they avoid reinitialisation and\nonly update the value to the reference of the signal<\/em>.<\/p>\n
This means we can also do some minor optimisations like avoiding rerendering\nof the DOM.<\/p>\n
At first, I didn't get it because it was an entirely new thing to me. After a bit\nof reading, I think I get the idea now if combined with useComputed<\/code>. To localise\nthe state into only a component, Preact suggests to use useComputed<\/code> with useSignal<\/code>.<\/p>\n
For example, prior to the use of useComputed<\/code>, I would access a value\ninside a JSX\/TSX and compare it to another value to render a JSX\/TSX element.<\/p>\n
    <<\/span>div<\/span>><\/span><\/span>\n     {<\/span>statuscode<\/span>.<\/span>value<\/span> !==<\/span> 200<\/span> &&<\/span> (<\/span><\/span>\n      <<\/span>a<\/span><\/span>\n        class<\/span>=<\/span>"<\/span>text-center rounded-md bg-white hover:bg-gray-100 text-black py-2 focus:outline-2 focus:outline-offset-2 focus:outline-white-500 <\/span>"<\/span><\/span>\n        href<\/span>=<\/span>"<\/span>\/signup<\/span>"<\/span><\/span>\n      ><\/span><\/span>\n        Create<\/span> Account<\/span><\/span>\n      <<\/span>\/<\/span>a<\/span>><\/span>}<\/span><\/span>\n      <<\/span>\/<\/span>div<\/span>><\/span><\/span>\n<\/span><\/code><\/pre>\n
After I knew about useComputed<\/code>, I just put it inside a useComputed<\/code> declaration<\/p>\n
  const<\/span> showSignup<\/span> =<\/span> useComputed<\/span>(<\/span>(<\/span>)<\/span> =><\/span><\/span>\n    statuscode<\/span>.<\/span>value<\/span> !==<\/span> 200<\/span> &&<\/span> (<\/span><\/span>\n      <<\/span>a<\/span><\/span>\n        class<\/span>=<\/span>"<\/span>text-center rounded-md bg-white hover:bg-gray-100 text-black py-2 focus:outline-2 focus:outline-offset-2 focus:outline-white-500 <\/span>"<\/span><\/span>\n        href<\/span>=<\/span>"<\/span>\/signup<\/span>"<\/span><\/span>\n      ><\/span><\/span>\n        Create<\/span> Account<\/span><\/span>\n      <<\/span>\/<\/span>a<\/span>><\/span><\/span>\n    )<\/span><\/span>\n  )<\/span>;<\/span><\/span><\/code><\/pre>\n
I won't claim that it actually does optimisations but this is what Preact said\nwhen accessing a signal's value inside a JSX.<\/p>\n
\n
In Preact, accessing a signal's .value property from within a component\nautomatically re-renders the component when that signal's value changes.<\/p>\n<\/blockquote>\n
Accessing the signal's value in an event handler does not cause a re-render.<\/p>\n
Once the value changes, only that part of the component updates, and not the\nwhole component itself.<\/p>\n
Here is an ASMR of the login and register. Of course I will use the\ntemplate's background gradient \ud83e\udd23<\/p>\n
How About the Backend?\u00a7<\/a\n>\n<\/h1>\n
I have to adjust some changes to the backend when experimenting the frontend.<\/p>\n
I noticed that I have to handle empty inputs as well which sends a 422 UNPROCESSABLE ENTITY<\/code> (looks\nlike it's called UNPROCESSABLE CONTENT<\/code> in the https:\/\/httpwg.org\/specs\/rfc9110.html#rfc.section.15.5.21<\/a> hmmmm).<\/p>\n
Other errors or status codes returned are 401 UNAUTHORIZED<\/code>, 500 INTERNAL SERVER ERROR<\/code>, 409 CONFLICT<\/code> and 400 BAD REQUEST<\/code>.<\/p>\n
I also made an API endpoint \/api\/auth\/profile<\/code> to check if the session token is still valid or not with the following code\nbelow.<\/p>\n
pub<\/span> async<\/span> fn<\/span> profile<\/span>(<\/span><\/span>\n    jar<\/span>:<\/span> CookieJar<\/span>,<\/span><\/span>\n)<\/span> -><\/span> anyhow<\/span>::<\/span>Result<\/span><<\/span>(<\/span>StatusCode<\/span>,<\/span> Json<\/span><<\/span>UserInfoResponse<\/span>><\/span>)<\/span>,<\/span> StatusCode<\/span>><\/span> {<\/span><\/span>\n    tracing<\/span>::<\/span>info!<\/span>(<\/span>?<\/span>jar<\/span>)<\/span>;<\/span><\/span>\n    let<\/span> Some<\/span>(<\/span>token<\/span>)<\/span> =<\/span> jar<\/span><\/span>\n        .<\/span>get<\/span>(<\/span>"<\/span>session_token<\/span>"<\/span>)<\/span><\/span>\n        .<\/span>map<\/span>(<\/span>|<\/span>cookie<\/span>|<\/span> cookie<\/span>.<\/span>value<\/span>(<\/span>)<\/span>.<\/span>to_owned<\/span>(<\/span>)<\/span>)<\/span><\/span>\n    else<\/span> {<\/span><\/span>\n        return<\/span> Err<\/span>(<\/span>StatusCode<\/span>::<\/span>UNAUTHORIZED<\/span>)<\/span>;<\/span><\/span>\n    }<\/span>;<\/span><\/span>\n    let<\/span> jwt_service<\/span> =<\/span> JwtService<\/span>::<\/span>new<\/span>(<\/span>"<\/span>some-secret<\/span>"<\/span>,<\/span> None<\/span>)<\/span>;<\/span><\/span>\n    let<\/span> claims<\/span> =<\/span> jwt_service<\/span><\/span>\n        .<\/span>verify<\/span>(<\/span>&<\/span>token<\/span>,<\/span> None<\/span>,<\/span> None<\/span>,<\/span> None<\/span>,<\/span> None<\/span>)<\/span><\/span>\n        .<\/span>map_err<\/span>(<\/span>|<\/span>err<\/span>|<\/span> {<\/span><\/span>\n            tracing<\/span>::<\/span>error!<\/span>(<\/span>%<\/span>err<\/span>)<\/span>;<\/span><\/span>\n            StatusCode<\/span>::<\/span>UNAUTHORIZED<\/span><\/span>\n        }<\/span>)<\/span>?<\/span>;<\/span><\/span>\n<\/span>\n    let<\/span> Some<\/span>(<\/span>id<\/span>)<\/span> =<\/span> claims<\/span>.<\/span>sub <\/span>else<\/span> {<\/span><\/span>\n        return<\/span> Err<\/span>(<\/span>StatusCode<\/span>::<\/span>UNAUTHORIZED<\/span>)<\/span>;<\/span><\/span>\n    }<\/span>;<\/span><\/span>\n    let<\/span> Some<\/span>(<\/span>email<\/span>)<\/span> =<\/span> claims<\/span>.<\/span>iss <\/span>else<\/span> {<\/span><\/span>\n        return<\/span> Err<\/span>(<\/span>StatusCode<\/span>::<\/span>UNAUTHORIZED<\/span>)<\/span>;<\/span><\/span>\n    }<\/span>;<\/span><\/span>\n<\/span>\n    let<\/span> mut<\/span> user_info<\/span> =<\/span> UserInfoResponse<\/span> {<\/span><\/span>\n        id<\/span>,<\/span><\/span>\n        email<\/span>,<\/span><\/span>\n        roles<\/span>:<\/span> Vec<\/span>::<\/span>new<\/span>(<\/span>)<\/span>,<\/span><\/span>\n        first_name<\/span>:<\/span> None<\/span>,<\/span><\/span>\n        last_name<\/span>:<\/span> None<\/span>,<\/span><\/span>\n    }<\/span>;<\/span><\/span>\n<\/span>\n    if<\/span> let<\/span> Some<\/span>(<\/span>extras<\/span>)<\/span> =<\/span> &<\/span>claims<\/span>.<\/span>extras <\/span>{<\/span><\/span>\n        let<\/span> first_name<\/span> =<\/span> extras<\/span>.<\/span>get<\/span>(<\/span>"<\/span>first_name<\/span>"<\/span>)<\/span>;<\/span><\/span>\n        let<\/span> last_name<\/span> =<\/span> extras<\/span>.<\/span>get<\/span>(<\/span>"<\/span>last_name<\/span>"<\/span>)<\/span>;<\/span><\/span>\n        let<\/span> roles_s<\/span> =<\/span> extras<\/span>.<\/span>get<\/span>(<\/span>"<\/span>roles<\/span>"<\/span>)<\/span>;<\/span><\/span>\n<\/span>\n        user_info<\/span>.<\/span>first_name <\/span>=<\/span> first_name<\/span>.<\/span>cloned<\/span>(<\/span>)<\/span>;<\/span><\/span>\n        user_info<\/span>.<\/span>last_name <\/span>=<\/span> last_name<\/span>.<\/span>cloned<\/span>(<\/span>)<\/span>;<\/span><\/span>\n<\/span>\n        if<\/span> let<\/span> Some<\/span>(<\/span>roles<\/span>)<\/span> =<\/span> roles_s<\/span> {<\/span><\/span>\n            let<\/span> role_str<\/span>:<\/span> Vec<\/span><<\/span>&<\/span>str<\/span>><\/span> =<\/span> roles<\/span>.<\/span>split<\/span>(<\/span>'<\/span>,<\/span>'<\/span>)<\/span>.<\/span>collect<\/span>(<\/span>)<\/span>;<\/span><\/span>\n            let<\/span> roles<\/span>:<\/span> Vec<\/span><<\/span>i32<\/span>><\/span> =<\/span> role_str<\/span><\/span>\n                .<\/span>iter<\/span>(<\/span>)<\/span><\/span>\n                .<\/span>map<\/span>(<\/span>|<\/span>x<\/span>|<\/span> x<\/span>.<\/span>parse<\/span>::<\/span><<\/span>i32<\/span>><\/span>(<\/span>)<\/span>)<\/span><\/span>\n                .<\/span>collect<\/span>::<\/span><<\/span>Result<\/span><<\/span>Vec<\/span><<\/span>i32<\/span>><\/span>,<\/span> ParseIntError<\/span>><\/span>><\/span>(<\/span>)<\/span><\/span>\n                .<\/span>map_err<\/span>(<\/span>|<\/span>err<\/span>|<\/span> {<\/span><\/span>\n                    tracing<\/span>::<\/span>error!<\/span>(<\/span>%<\/span>err<\/span>)<\/span>;<\/span><\/span>\n                    StatusCode<\/span>::<\/span>UNAUTHORIZED<\/span><\/span>\n                }<\/span>)<\/span>?<\/span>;<\/span><\/span>\n            user_info<\/span>.<\/span>roles <\/span>=<\/span> roles<\/span>;<\/span><\/span>\n        }<\/span><\/span>\n    }<\/span>;<\/span><\/span>\n    tracing<\/span>::<\/span>info!<\/span>(<\/span>"<\/span>Passed validation \u2705<\/span>"<\/span>)<\/span>;<\/span><\/span>\n<\/span>\n    Ok<\/span>(<\/span>(<\/span>StatusCode<\/span>::<\/span>OK<\/span>,<\/span> Json<\/span>(<\/span>user_info<\/span>)<\/span>)<\/span>)<\/span><\/span>\n}<\/span><\/span><\/code><\/pre>\n
I am pretty sure the logic can be reused as middleware for\nhandling routes that require a valid session token.<\/p>\n
Security-wise, I would probably ask an expert or do research\nbecause I believe if someone has the token string taken from\nthe cookie, it would be so problematic. The only protection it\nhas is the signature verification of which I am unsure if it\nis very protective enough.<\/p>\n
That's all. I'll continue building the database crate again\nfor the products schema.<\/p>\n"},{"title":"Part 4: So I am building a fullstack project","published":"2026-03-12T01:55:47+00:00","updated":"2026-03-12T01:55:47+00:00","author":{"name":"\n            Soc Virnyl Estela\n          "},"link":{"@attributes":{"rel":"alternate","type":"text\/html","href":"https:\/\/uncomfyhalomacro.pl\/blog\/202603095442\/"}},"id":"https:\/\/uncomfyhalomacro.pl\/blog\/202603095442\/","content":"
What Was Done So Far?\u00a7<\/a\n>\n<\/h1>\n
I have implemented the initial JSON Web Token logic for the auth<\/code> crate.<\/p>\n
Using the builder pattern, I came up with the following builder struct and its\nassociated Claims struct to generate claims (or payload) for generating the\ntoken.<\/p>\n
You can see how it was used here in the following lines of\ncode, alongside the JwtService struct<\/a>\nfrom our service.rs<\/code>.<\/p>\n
It's flexible enough that I can basically tell it's just a wrapper around the jsonwebtoken<\/code> crate that it is using.<\/p>\n
    fn<\/span> test_correct_signature<\/span>(<\/span>)<\/span> {<\/span><\/span>\n        let<\/span> secret<\/span> =<\/span> "<\/span>dumb_secret<\/span>"<\/span>;<\/span><\/span>\n        let<\/span> builder<\/span> =<\/span> ClaimsBuilder<\/span>::<\/span>new<\/span>(<\/span>5<\/span> *<\/span> 60<\/span>,<\/span> secret<\/span>.<\/span>to_string<\/span>(<\/span>)<\/span>)<\/span>;<\/span><\/span>\n        let<\/span> claims<\/span> =<\/span> builder<\/span>.<\/span>build<\/span>(<\/span>)<\/span>;<\/span><\/span>\n        let<\/span> jwt_service<\/span> =<\/span> JwtService<\/span>::<\/span>new<\/span>(<\/span>secret<\/span>,<\/span> None<\/span>)<\/span>;<\/span><\/span>\n        let<\/span> token_res<\/span> =<\/span> jwt_service<\/span>.<\/span>sign<\/span>(<\/span>&<\/span>claims<\/span>)<\/span>;<\/span><\/span>\n        assert!<\/span>(<\/span>token_res<\/span>.<\/span>is_ok<\/span>(<\/span>)<\/span>)<\/span>;<\/span><\/span>\n        let<\/span> token<\/span> =<\/span> token_res<\/span>.<\/span>unwrap<\/span>(<\/span>)<\/span>;<\/span><\/span>\n        assert!<\/span>(<\/span>jwt_service<\/span>.<\/span>verify<\/span>(<\/span>&<\/span>token<\/span>,<\/span> None<\/span>,<\/span> None<\/span>,<\/span> None<\/span>,<\/span> None<\/span>)<\/span>.<\/span>is_ok<\/span>(<\/span>)<\/span>)<\/span>;<\/span><\/span>\n    }<\/span><\/span><\/code><\/pre>
Fixing Some SQL Schema and Logic\u00a7<\/a\n>\n<\/h1>\n
So far the just<\/code> recipes for running the tests are problematic (and they still are)\nbecause I find bugs here and there.<\/p>\n
What I realise is because the migrations to create the tables should have IF NOT EXISTS<\/code>\nso it would only create the tables if they do not exist.<\/p>\n
Another issue is the lack of variables that are related to the tests themselves. For now,\nthe current solution is just a band-aide. I am pretty surer there is a better way to\nstructure the tests like putting a separate justfile script in a separate directory.<\/p>\n
But for now, I'll just let it be.<\/p>\n
The next step is adding Oath2.0 logic using openidconnect<\/code> crate to the auth<\/code> crate.<\/p>\n
Once that's done, I'll try implementing the payments<\/code> crate. I believe there is already\nclient libraries for Payrex and Stripe payments. Not sure about Wise, I can try Wise too\nsince I already have a Wise card.<\/p>\n
So far so good. No progress on the frontend yet.<\/p>\n"},{"title":"Part 3: So I am building a fullstack project","published":"2026-03-10T00:51:55+00:00","updated":"2026-03-10T00:51:55+00:00","author":{"name":"\n            \n              Unknown\n            \n          "},"link":{"@attributes":{"rel":"alternate","type":"text\/html","href":"https:\/\/uncomfyhalomacro.pl\/blog\/202603085047\/"}},"id":"https:\/\/uncomfyhalomacro.pl\/blog\/202603085047\/","content":"
Now back to the progress. I was busy with other personal stuff lately.<\/p>\n
I've decided to rework some of the SQL schemas of the tables.<\/p>\n
CREATE<\/span> TABLE<\/span> users<\/span> (<\/span><\/span>\n    id uuid <\/span>PRIMARY KEY<\/span>,<\/span><\/span>\n    roles <\/span>INTEGER<\/span>[]<\/span> NOT NULL<\/span>,<\/span><\/span>\n    first_name <\/span>TEXT<\/span> DEFAULT<\/span> NULL<\/span>,<\/span><\/span>\n    last_name <\/span>TEXT<\/span> DEFAULT<\/span> NULL<\/span>,<\/span><\/span>\n    email <\/span>TEXT<\/span> NOT NULL<\/span>,<\/span><\/span>\n    created_at <\/span>TIMESTAMP<\/span>TZ<\/span> DEFAULT<\/span> NOW<\/span>(<\/span>)<\/span>,<\/span><\/span>\n    updated_at <\/span>TIMESTAMP<\/span>TZ<\/span> DEFAULT<\/span> NOW<\/span>(<\/span>)<\/span>,<\/span><\/span>\n    CONSTRAINT<\/span> unique_email <\/span>UNIQUE<\/span> (email)<\/span><\/span>\n);<\/span><\/span>\n<\/span>\nCREATE<\/span> TABLE<\/span> user_passwords<\/span> (<\/span><\/span>\n    id uuid <\/span>PRIMARY KEY<\/span>,<\/span><\/span>\n    user_id uuid <\/span>NOT NULL<\/span> REFERENCES<\/span> users(id) <\/span>ON DELETE CASCADE<\/span>,<\/span><\/span>\n    password_hash <\/span>TEXT<\/span> NOT NULL<\/span>,<\/span><\/span>\n    created_at <\/span>TIMESTAMP<\/span>TZ<\/span> DEFAULT<\/span> NOW<\/span>(<\/span>)<\/span>,<\/span><\/span>\n    updated_at <\/span>TIMESTAMP<\/span>TZ<\/span> DEFAULT<\/span> NOW<\/span>(<\/span>)<\/span>,<\/span><\/span>\n    CONSTRAINT<\/span> FK_userId <\/span>FOREIGN KEY<\/span> (user_id) <\/span>REFERENCES<\/span> users(id),<\/span><\/span>\n    CONSTRAINT<\/span> unique_user_id <\/span>UNIQUE<\/span> (user_id)<\/span><\/span>\n);<\/span><\/span>\n<\/span>\nCREATE<\/span> TABLE<\/span> audit_logs<\/span> (<\/span><\/span>\n    id UUID <\/span>NOT NULL<\/span>,<\/span><\/span>\n    table_oid <\/span>OID<\/span>, <\/span>--<\/span> pertains to the oid of the table in the information schema. aliased to a u32 integer<\/span><\/span>\n    entity_id UUID <\/span>NOT NULL<\/span>, <\/span>--<\/span> this pertains to the table row's id<\/span><\/span>\n    event_type <\/span>TEXT<\/span> NOT NULL<\/span>, <\/span>--<\/span> actions that happened such as adding products in stock. Removing them will remove this row as well<\/span><\/span>\n    data<\/span> JSONB,<\/span><\/span>\n    created_at <\/span>TIMESTAMP<\/span>TZ<\/span> DEFAULT<\/span> now<\/span>(<\/span>)<\/span><\/span>\n);<\/span><\/span>\n<\/span>\nCREATE<\/span> TABLE<\/span> products<\/span> (<\/span><\/span>\n    id UUID <\/span>PRIMARY KEY<\/span>,<\/span><\/span>\n    seller_id UUID <\/span>NOT NULL<\/span>,<\/span><\/span>\n    name<\/span> TEXT<\/span> NOT NULL<\/span>,<\/span><\/span>\n    description<\/span> TEXT<\/span>,<\/span><\/span>\n    item_type <\/span>TEXT<\/span> NOT NULL<\/span>, <\/span>--<\/span> product, service<\/span><\/span>\n    unit_price <\/span>NUMERIC<\/span>(<\/span>12<\/span>,<\/span>2<\/span>)<\/span>,<\/span><\/span>\n    currency <\/span>CHAR<\/span>(<\/span>3<\/span>)<\/span>,<\/span><\/span>\n    metadata JSONB, <\/span>--<\/span> extra metadata the seller or us wants to include without changing the table<\/span><\/span>\n    is_active <\/span>BOOLEAN<\/span> DEFAULT<\/span> TRUE,<\/span><\/span>\n    created_at <\/span>TIMESTAMP<\/span>TZ<\/span> DEFAULT<\/span> now<\/span>(<\/span>)<\/span>,<\/span><\/span>\n    updated_at <\/span>TIMESTAMP<\/span>TZ<\/span> DEFAULT<\/span> now<\/span>(<\/span>)<\/span><\/span>\n);<\/span><\/span><\/code><\/pre>\n
I already made some migration scripts as well, except for products (I actually\nhave forgotten I didn't implement a migration for products).<\/p>\n
Switching over to Nix for Development Environment Management\u00a7<\/a\n>\n<\/h1>\n
I made some changes as well on how I manage my development environment. I am\nusing nix<\/code>, specifically the one made by https:\/\/determinate.systems<\/a> since\nthey have the best configuration on a macOS environment as far as I know. Since\nI don't have time to relearn the Nix language, I decided to use what I can\nremember plus with the help of LLMs like ChatGPT and Gemini.<\/p>\n
I also adjusted my shell environment with the help of direnv<\/code>. Mmfallacy<\/a> suggested\nto add some sort of indicator to my shell prompt that I am inside a Nix environment, so I came up with\nmodifying the following lines to my ~\/.zshrc<\/code>.<\/p>\n
export<\/span> PROMPT<\/span>=<\/span>$'<\/span>\\$vcs_info_msg_0_%F{010} %B%F{blue}%n%b%f in %F{black}%K{7}<%3~>%k%f %F{yellow}[%y] %F{green}@ %B%U%F{white}%M%u%b %F{red}%T %F{white}%D{%b %d %Y (%A)}<\/span><\/span>\n%{\\x1b[0m%}%(?.%F{green} .%F{red}%? )%%%F{none} <\/span>'<\/span><\/span>\nexport<\/span> PROMPT<\/span>=<\/span>'<\/span>%F{#2685bd}${DIRENV_PROMPT_PREFIX:+$DIRENV_PROMPT_PREFIX}%f<\/span>'<\/span>$<\/span>PROMPT<\/span><\/span><\/code><\/pre>\n
To get the DIRENV_PROMPT_PREFIX<\/code> environment variable, I adjusted the project's .envrc<\/code> that direnv<\/code> uses<\/p>\n
use flake<\/span><\/span>\n<\/span>\ndescription=$(nix flake metadata --json | jq '.description' | tr -d '"')<\/span><\/span>\nfingerprint=$(nix flake metadata --json | jq '.fingerprint' | tr -d '"' | cut -c1-6)<\/span><\/span>\nexport DIRENV_PROMPT_PREFIX="\u2744\ufe0f(nix-${description}-${fingerprint})<\/span><\/span>\n"<\/span><\/span><\/code><\/pre>\n
This way, direnv<\/code> will inject new environmental variables to my shell's RC. Now, my prompt looks like\nthe screenshot below.<\/p>\n
\"screenshot<\/p>\n
I am not sure how useful description<\/code> and fingerprint<\/code> fields are but I guess\nthat's fine because I can determine which environment using their uniqueness whenever if I have multiple\nprojects managed by flakes in the future.<\/p>\n
Because of this, even I tested that it's possible to use PostgreSQL outside homebrew so I uninstalled it and\nadjusted my justfile<\/code> now to look like this<\/p>\n
set shell := ["bash", "-cu"]<\/span><\/span>\nset dotenv-load<\/span><\/span>\n<\/span>\nserve: serve<\/span><\/span>\ntest-user-models:<\/span><\/span>\n\tjust migrate-down || true<\/span><\/span>\n\tjust migrate-up || true<\/span><\/span>\n\tcargo test<\/span><\/span>\n<\/span>\nmigrate-up:<\/span><\/span>\n\tcargo run --bin up_0001_users || true<\/span><\/span>\n\tcargo run --bin up_0002_audit_logs<\/span><\/span>\n<\/span>\nmigrate-down:<\/span><\/span>\n\tcargo run --bin down_0001_users || true<\/span><\/span>\n\tcargo run --bin down_0002_audit_logs<\/span><\/span>\n<\/span>\n<\/span>\nbuild-backend:<\/span><\/span>\n\tcargo build<\/span><\/span>\n<\/span>\nserve-backend: build-backend<\/span><\/span>\n\tcargo run<\/span><\/span>\n<\/span>\nbuild-frontend:<\/span><\/span>\n\tcd frontend && deno task build<\/span><\/span>\n<\/span>\nserve-frontend:<\/span><\/span>\n\tcd frontend && deno task dev<\/span><\/span>\n<\/span>\nstart-db:<\/span><\/span>\n\tif [ ! -d "${PGDATA}" ]; then initdb -D ${PGDATA}; fi<\/span><\/span>\n\tpg_ctl -o "${INIT_DB_OPTIONS}" -D ${PGDATA} start<\/span><\/span>\n<\/span>\nstop-db:<\/span><\/span>\n\tpg_ctl -D ${PGDATA} stop<\/span><\/span>\n<\/span><\/code><\/pre>
Using Something Similar to Builder Pattern\u00a7<\/a\n>\n<\/h1>\n
I am also using the builder pattern when I am creating new methods for each of my model structs.<\/p>\n
For example, I made a separate method for save_to_db<\/code> because I believe that it should be a separate concern.<\/p>\n
As demonstration, here is the Users<\/code> model<\/p>\n
impl<\/span> Users<\/span> {<\/span><\/span>\n    pub<\/span> fn<\/span> new<\/span>(<\/span>first_name<\/span>:<\/span> Option<\/span><<\/span>String<\/span>><\/span>,<\/span> last_name<\/span>:<\/span> Option<\/span><<\/span>String<\/span>><\/span>,<\/span> email<\/span>:<\/span> String<\/span>)<\/span> -><\/span> Self<\/span> {<\/span><\/span>\n        Self<\/span> {<\/span><\/span>\n            id<\/span>:<\/span> Uuid<\/span>::<\/span>new_v4<\/span>(<\/span>)<\/span>,<\/span><\/span>\n            roles<\/span>:<\/span> Vec<\/span>::<\/span>new<\/span>(<\/span>)<\/span>,<\/span><\/span>\n            first_name<\/span>,<\/span><\/span>\n            last_name<\/span>,<\/span><\/span>\n            email<\/span>,<\/span><\/span>\n            created_at<\/span>:<\/span> Utc<\/span>::<\/span>now<\/span>(<\/span>)<\/span>,<\/span><\/span>\n            updated_at<\/span>:<\/span> Utc<\/span>::<\/span>now<\/span>(<\/span>)<\/span>,<\/span><\/span>\n        }<\/span><\/span>\n    }<\/span><\/span>\n<\/span>\n    pub<\/span> async<\/span> fn<\/span> save_to_db<\/span>(<\/span>&<\/span>self<\/span>,<\/span> pool<\/span>:<\/span> &<\/span>Pool<\/span>)<\/span> -><\/span> Result<\/span><<\/span>Row<\/span>,<\/span> PoolError<\/span>><\/span> {<\/span><\/span>\n        let<\/span> client<\/span> =<\/span> pool<\/span>.<\/span>get<\/span>(<\/span>)<\/span>.<\/span>await<\/span>?<\/span>;<\/span><\/span>\n        let<\/span> stmt<\/span> =<\/span> client<\/span>.<\/span>prepare_cached<\/span>(<\/span>"<\/span>INSERT INTO users (id, roles, first_name, last_name, email, created_at, updated_at) VALUES ($1, $2, $3, $4, $5, $6, $7) RETURNING *<\/span>"<\/span>)<\/span>.<\/span>await<\/span>?<\/span>;<\/span><\/span>\n        let<\/span> roles<\/span>:<\/span> Vec<\/span><<\/span>i32<\/span>><\/span> =<\/span> self<\/span>.<\/span>roles<\/span>.<\/span>iter<\/span>(<\/span>)<\/span>.<\/span>map<\/span>(<\/span>|<\/span>&<\/span>x<\/span>|<\/span> x<\/span> as<\/span> i32<\/span>)<\/span>.<\/span>collect<\/span>(<\/span>)<\/span>;<\/span><\/span>\n        let<\/span> row<\/span> =<\/span> client<\/span><\/span>\n            .<\/span>query_one<\/span>(<\/span><\/span>\n                &<\/span>stmt<\/span>,<\/span><\/span>\n                &<\/span>[<\/span><\/span>\n                    &<\/span>self<\/span>.<\/span>id<\/span>,<\/span><\/span>\n                    &<\/span>roles<\/span>,<\/span><\/span>\n                    &<\/span>self<\/span>.<\/span>first_name<\/span>,<\/span><\/span>\n                    &<\/span>self<\/span>.<\/span>last_name<\/span>,<\/span><\/span>\n                    &<\/span>self<\/span>.<\/span>email<\/span>,<\/span><\/span>\n                    &<\/span>self<\/span>.<\/span>created_at<\/span>,<\/span><\/span>\n                    &<\/span>self<\/span>.<\/span>updated_at<\/span>,<\/span><\/span>\n                ]<\/span>,<\/span><\/span>\n            )<\/span><\/span>\n            .<\/span>await<\/span>?<\/span>;<\/span><\/span>\n        Ok<\/span>(<\/span>row<\/span>)<\/span><\/span>\n    }<\/span><\/span>\n<\/span>\n    pub<\/span> fn<\/span> hash_password<\/span>(<\/span><\/span>\n        &<\/span>self<\/span>,<\/span><\/span>\n        plain_text_password<\/span>:<\/span> &<\/span>str<\/span>,<\/span><\/span>\n    )<\/span> -><\/span> Result<\/span><<\/span>UserPasswords<\/span>,<\/span> password_hash<\/span>::<\/span>Error<\/span>><\/span> {<\/span><\/span>\n        let<\/span> salt<\/span> =<\/span> SaltString<\/span>::<\/span>generate<\/span>(<\/span>&<\/span>mut<\/span> OsRng<\/span>)<\/span>;<\/span><\/span>\n        let<\/span> password<\/span> =<\/span> plain_text_password<\/span>.<\/span>as_bytes<\/span>(<\/span>)<\/span>;<\/span><\/span>\n        let<\/span> argon2<\/span> =<\/span> Argon2<\/span>::<\/span>default<\/span>(<\/span>)<\/span>;<\/span><\/span>\n        let<\/span> password_hash<\/span> =<\/span> argon2<\/span>.<\/span>hash_password<\/span>(<\/span>password<\/span>,<\/span> &<\/span>salt<\/span>)<\/span>?<\/span>.<\/span>to_string<\/span>(<\/span>)<\/span>;<\/span><\/span>\n        let<\/span> id<\/span> =<\/span> Uuid<\/span>::<\/span>new_v4<\/span>(<\/span>)<\/span>;<\/span><\/span>\n        Ok<\/span>(<\/span>UserPasswords<\/span> {<\/span><\/span>\n            id<\/span>,<\/span><\/span>\n            user_id<\/span>:<\/span> self<\/span>.<\/span>id<\/span>,<\/span><\/span>\n            password_hash<\/span>,<\/span><\/span>\n            created_at<\/span>:<\/span> Utc<\/span>::<\/span>now<\/span>(<\/span>)<\/span>,<\/span><\/span>\n            updated_at<\/span>:<\/span> Utc<\/span>::<\/span>now<\/span>(<\/span>)<\/span>,<\/span><\/span>\n        }<\/span>)<\/span><\/span>\n    }<\/span><\/span>\n<\/span>\n    pub<\/span> async<\/span> fn<\/span> get_user_password_hash<\/span>(<\/span>&<\/span>self<\/span>,<\/span> pool<\/span>:<\/span> &<\/span>Pool<\/span>)<\/span> -><\/span> Result<\/span><<\/span>UserPasswords<\/span>,<\/span> PoolError<\/span>><\/span> {<\/span><\/span>\n        UserPasswords<\/span>::<\/span>from_user_id<\/span>(<\/span>pool<\/span>,<\/span> &<\/span>self<\/span>.<\/span>id<\/span>)<\/span>.<\/span>await<\/span><\/span>\n    }<\/span><\/span>\n}<\/span><\/span><\/code><\/pre>\n
And the tests that go alongside it.<\/p>\n
\/\/<\/span> snipped...<\/span><\/span>\n    #<\/span>[<\/span>tokio<\/span>::<\/span>test<\/span>]<\/span><\/span>\n    async<\/span> fn<\/span> add_user<\/span>(<\/span>)<\/span> {<\/span><\/span>\n        dotenv<\/span>(<\/span>)<\/span>.<\/span>ok<\/span>(<\/span>)<\/span>;<\/span><\/span>\n        let<\/span> cfg<\/span> =<\/span> Config<\/span>::<\/span>from_env<\/span>(<\/span>)<\/span>.<\/span>unwrap<\/span>(<\/span>)<\/span>;<\/span><\/span>\n        let<\/span> pool<\/span> =<\/span> cfg<\/span>.<\/span>pg<\/span>.<\/span>create_pool<\/span>(<\/span>Some<\/span>(<\/span>Runtime<\/span>::<\/span>Tokio1<\/span>)<\/span>,<\/span> NoTls<\/span>)<\/span>.<\/span>unwrap<\/span>(<\/span>)<\/span>;<\/span><\/span>\n        assert!<\/span>(<\/span>table_exists<\/span>(<\/span>&<\/span>pool<\/span>,<\/span> "<\/span>users<\/span>"<\/span>)<\/span>.<\/span>await<\/span>.<\/span>unwrap<\/span>(<\/span>)<\/span>)<\/span>;<\/span><\/span>\n        let<\/span> new_user<\/span> =<\/span> Users<\/span>::<\/span>new<\/span>(<\/span><\/span>\n            Some<\/span>(<\/span>"<\/span>David<\/span>"<\/span>.<\/span>to_string<\/span>(<\/span>)<\/span>)<\/span>,<\/span><\/span>\n            Some<\/span>(<\/span>"<\/span>Attenborough<\/span>"<\/span>.<\/span>to_string<\/span>(<\/span>)<\/span>)<\/span>,<\/span><\/span>\n            "<\/span>contact@david_attenborough.com<\/span>"<\/span>.<\/span>to_string<\/span>(<\/span>)<\/span>,<\/span><\/span>\n        )<\/span>;<\/span><\/span>\n        let<\/span> secure_password<\/span> =<\/span> "<\/span>very_secure<\/span>"<\/span>;<\/span><\/span>\n<\/span>\n        new_user<\/span>.<\/span>save_to_db<\/span>(<\/span>&<\/span>pool<\/span>)<\/span>.<\/span>await<\/span>.<\/span>unwrap<\/span>(<\/span>)<\/span>;<\/span><\/span>\n        let<\/span> password<\/span> =<\/span> new_user<\/span>.<\/span>hash_password<\/span>(<\/span>secure_password<\/span>)<\/span>.<\/span>unwrap<\/span>(<\/span>)<\/span>;<\/span><\/span>\n        assert!<\/span>(<\/span>password<\/span>.<\/span>save_to_db<\/span>(<\/span>&<\/span>pool<\/span>)<\/span>.<\/span>await<\/span>.<\/span>is_ok<\/span>(<\/span>)<\/span>)<\/span>;<\/span><\/span>\n    }<\/span><\/span>\n<\/span>\n    #<\/span>[<\/span>tokio<\/span>::<\/span>test<\/span>]<\/span><\/span>\n    async<\/span> fn<\/span> verify_password<\/span>(<\/span>)<\/span> {<\/span><\/span>\n        dotenv<\/span>(<\/span>)<\/span>.<\/span>ok<\/span>(<\/span>)<\/span>;<\/span><\/span>\n        let<\/span> cfg<\/span> =<\/span> Config<\/span>::<\/span>from_env<\/span>(<\/span>)<\/span>.<\/span>unwrap<\/span>(<\/span>)<\/span>;<\/span><\/span>\n        let<\/span> pool<\/span> =<\/span> cfg<\/span>.<\/span>pg<\/span>.<\/span>create_pool<\/span>(<\/span>Some<\/span>(<\/span>Runtime<\/span>::<\/span>Tokio1<\/span>)<\/span>,<\/span> NoTls<\/span>)<\/span>.<\/span>unwrap<\/span>(<\/span>)<\/span>;<\/span><\/span>\n        assert!<\/span>(<\/span>table_exists<\/span>(<\/span>&<\/span>pool<\/span>,<\/span> "<\/span>users<\/span>"<\/span>)<\/span>.<\/span>await<\/span>.<\/span>unwrap<\/span>(<\/span>)<\/span>)<\/span>;<\/span><\/span>\n        let<\/span> new_user<\/span> =<\/span> Users<\/span>::<\/span>new<\/span>(<\/span><\/span>\n            Some<\/span>(<\/span>"<\/span>Fred<\/span>"<\/span>.<\/span>to_string<\/span>(<\/span>)<\/span>)<\/span>,<\/span><\/span>\n            Some<\/span>(<\/span>"<\/span>Adila<\/span>"<\/span>.<\/span>to_string<\/span>(<\/span>)<\/span>)<\/span>,<\/span><\/span>\n            "<\/span>contact@fredadila.com<\/span>"<\/span>.<\/span>to_string<\/span>(<\/span>)<\/span>,<\/span><\/span>\n        )<\/span>;<\/span><\/span>\n        let<\/span> secure_password<\/span> =<\/span> "<\/span>very_secure<\/span>"<\/span>;<\/span><\/span>\n<\/span>\n        new_user<\/span>.<\/span>save_to_db<\/span>(<\/span>&<\/span>pool<\/span>)<\/span>.<\/span>await<\/span>.<\/span>unwrap<\/span>(<\/span>)<\/span>;<\/span><\/span>\n        let<\/span> password<\/span> =<\/span> new_user<\/span>.<\/span>hash_password<\/span>(<\/span>secure_password<\/span>)<\/span>.<\/span>unwrap<\/span>(<\/span>)<\/span>;<\/span><\/span>\n        assert!<\/span>(<\/span>password<\/span>.<\/span>save_to_db<\/span>(<\/span>&<\/span>pool<\/span>)<\/span>.<\/span>await<\/span>.<\/span>is_ok<\/span>(<\/span>)<\/span>)<\/span>;<\/span><\/span>\n        assert!<\/span>(<\/span>password<\/span>.<\/span>verify_password<\/span>(<\/span>"<\/span>incorrect<\/span>"<\/span>)<\/span>.<\/span>is_err<\/span>(<\/span>)<\/span>)<\/span>;<\/span><\/span>\n        assert!<\/span>(<\/span>password<\/span>.<\/span>verify_password<\/span>(<\/span>"<\/span>very_secure<\/span>"<\/span>)<\/span>.<\/span>is_ok<\/span>(<\/span>)<\/span>)<\/span>;<\/span><\/span>\n    }<\/span><\/span>\n\/\/<\/span> snipped...<\/span><\/span><\/code><\/pre>\n
That means if we ever write something from initializing new data to saving it to a database, it would look like<\/p>\n
let<\/span> password_plain<\/span> =<\/span> "<\/span>verysecure_password<\/span>"<\/span>;<\/span><\/span>\nlet<\/span> user<\/span> =<\/span> Users<\/span>::<\/span>new<\/span>(<\/span>somedata<\/span>)<\/span>;<\/span>  \/\/<\/span> somedata is just placeholder<\/span><\/span>\nlet<\/span> user_password<\/span> =<\/span> user<\/span>.<\/span>hash_password<\/span>(<\/span>password_plain<\/span>)<\/span>;<\/span><\/span>\nuser<\/span>.<\/span>save_to_db<\/span>(<\/span>&<\/span>pool<\/span>)<\/span>.<\/span>await<\/span>.<\/span>unwrap<\/span>(<\/span>)<\/span>;<\/span><\/span>\nuser_password<\/span>.<\/span>save_to_db<\/span>(<\/span>&<\/span>pool<\/span>)<\/span>.<\/span>await<\/span>.<\/span>unwrap<\/span>(<\/span>)<\/span>;<\/span><\/span><\/code><\/pre>\n
I believe I can wrap deadpool_postgres::Pool<\/code> with a Mutex<\/code> or Arc<\/code> because I\nfeel like I am repeating myself whenever I am passing a Pool<\/code> to the method.<\/p>\n
The reason that I am against it is because of the nature of the struct fields\nthemselves.<\/p>\n
They are clearly mirrored to the SQL schema that I made so if I\nwill add a pool<\/code> field, it would lose its representation and that will likely\nnot communicate well with the model I made.<\/p>\n
Hence, I am leaving it at that, to ensure they don't lose what they represent. But I think I can wrap the pool<\/code> outside the structs themselves<\/em>.<\/p>\n
What's next\u00a7<\/a\n>\n<\/h1>\n
I will continue adding the migration logic for products and proceed building the auth<\/code> crate. The auth<\/code> crate should have the following<\/p>\n
Another project update for our mercado-series<\/a>.<\/p>\n
I also encountered a bug<\/a> in utoipa\nof which is not very obvious at first. The workaround is ensuring that functions with\nthe same paths have different namings.<\/p>\n