shim: query CLI

[pepyakin]

This changeset introduces a command allowing querying the sugondat from the CLI. At this time it only adds submit functionality: submit a blob to a given namespace.

As such this change is more about laying framework.

Builds on top of #32.

[rphmeier]

these are query parameters for the DA layer block / number, right?

[pepyakin]

yep! generally shim doesn't know anything about rollup blocks. Only DA blocks and blobs.

[rphmeier]

Looks good on top of #32

[rphmeier]

Is this creating a default key?

[rphmeier]

https://docs.rs/subxt/latest/subxt/tx/struct.TxClient.html#method.create_signed

I don't quite understand what's happening here and what the type of the created Signer is.

[pepyakin]

create_signed + the next thing is actually inlining this function

https://github.com/thrumdev/sugondat/blob/16bf54cdc99dfdbafae64b5fb18168310807f56e/sugondat-da-adapter/src/service.rs#L214-L220

I did this in an attempt to make the error message a bit less confusing. I have no idea what other_data (specified by Default::default) is and why it is used. The signer key is passed by from.

[rphmeier]

Okay, makes sense. I missed the from parameter somehow.

The other_data would be SignedExtra stuff, I guess, which might be desired for mortality in particular.

[gabriele-0201]

another possible implementation would be:

fn read_namespace(namespace: &str) -> anyhow::Result<sugondat_nmt::Namespace> {

    if let Some(hex) = namespace.strip_prefix("0x") {
        let namespace = hex::decode(hex)?;
        let namespace: [u8; 4] = namespace.try_into().map_err(|e: Vec<u8>| {
            anyhow::anyhow!("namespace must be 4 bytes long, but was {}", e.len())
        })?;
        Ok(sugondat_nmt::Namespace::from_raw_bytes(namespace))
    }

    let namespace_id = namespace
        .parse::<u32>()
        .with_context(|| format!("cannot parse namespace id '{}'", namespace))?;
    Ok(sugondat_nmt::Namespace::with_namespace_id(namespace_id))
}

or cleaner but technically more computation required because the from_str_radix would perform an useless translation to u32

fn read_namespace(mut namespace: &str) -> anyhow::Result<sugondat_nmt::Namespace> {

    let namespace_id = match namespace.strip_prefix("0x") {
        Some(hex) => u32::from_str_radix(hex, 16).with_context(|| format!("cannot parse namespace id '{}'", namespace))?,
        None => namespace.parse::<u32>().with_context(|| format!("cannot parse namespace id '{}'", namespace))?
    };

    Ok(sugondat_nmt::Namespace::with_namespace_id(namespace_id))
}

(didn't test the error handling)

[pepyakin]

Hah, this is a funny one.

I specifically went for the implementation that you see here to disallow passing 0x0x0011, because those functions are not equivalent: one removes the occurance once and the other removes all occurances. However, I mixed them up and did the opposite what I inteded.

That said, I don't think this matters much and it's OK either way. Feel free to file a PR with the fix. I actually prefer the first option: for one, it's not unlikely that we change the size of the namespace #12 . Note that you it seems you are missing a return though.