The SubCommand enum in the main.rs file uses an attribute to name each command. Since Clap already converts the enum variant name to kebab case for the command name, this attribute is only required if a different name is required.

There are no examples in this enum where the manually configured name doesn’t match the generated one.

I didn’t know that. Do you want to push a change to this discussion?

pijul push danieleades@nest.pijul.com:pijul/pijul --to-channel :136

i would love to. As soon as I can get any release of pijul to clone, build, or install I will be happy to do so!

I got this far by using curl to download a tarball from crates.io

Can you use cargo install pijul --version 1.0.0-alpha.9?

i had tried alphas 5-8, but it seems 9 was the magic number!

• unable to clone using ssh - key is rejected while cloning, and it’s asking for a password.
• was able to clone using http
• i can’t record the changes (pijul record)- it claims there’s a syntax error in the generated file when I try to save it

any ideas?

incidentally, I can’t exit the file without saving either, I get stuck in a loop where I either have to kill the terminal or manually delete every line in the file before re-saving.

i see also that you have-

    /// Unrecords all changes upto the given hash
    #[clap(alias = "unrec", alias = "un")]
    Unrecord(Unrecord),

would you consider instead using the Clap setting InferSubcommands? I think this might do what you’re after

I’ve also noticed that pijul -h, pijul --help, and pijul help don’t work in this release

i think this block here might be breaking it?-

    for arg in std::env::args().skip(1) {
        if cmd.is_empty() {
            if app
                .get_subcommands()
                .any(|sub| sub.get_name() == arg || sub.get_all_aliases().any(|al| al == arg))
            {
                return Ok(false);
            }
            cmd.push_str("pijul-");
            cmd.push_str(&arg)
        } else {
            args.push(arg)
        }
    }

also these app settings-

#[clap(version = crate_version!(), author = crate_authors!(),...]

shouldn’t be required, as these are each the default behaviour if they’re not supplied.

at least that’s the case with structopt, i’ve not actually used the Clap beta, but I can’t imagine why they would have changed the behaviour.

I can’t confirm locally, since the help string seems to be broken

managed to record changes. now attempting to push them.

• pijul doesn’t like using SSH. Have tried generating a new key
• nest website crashes when I attempt to save the new public key

There was a problem with SSH keys, I just fixed it.

Thanks! And sorry again about the hiccups of the Nest.

the “use ‘InferSubcommands’ instead of aliases” patch might be slightly controversial as this could interfere with external subcommands (depending on how this is being used)

Thanks! And sorry again about the hiccups of the Nest.

no, thankyou for the quick fixes! I think there’s a lot of us out here who really want this project to succeed.

It’s ok. If we let external subcommands use names that look like shortcuts to standard commands, things might be super confusing (I can already imagine some StackOverflow questions).

would you consider using the native external subcommand support instead of the hand-rolled one?

https://docs.rs/structopt/0.3.20/structopt/#external-subcommands

Yes, absolutely. I tried to look for this in Clap’s documentation, and couldn’t find it.

clap’s documentation is pretty terrible, but you can use the structopt docs as a reference

Thanks!

i’ve got a feeling that the crate version and author are not pulled by default in Clap anymore. I think maybe you were right to add these manually…

Alright, I unrecorded the change (QRZ…) from the main channel, but change XJY… depended on it, so I had to unrecord it too.

Would you mind either:

• unrecording XJY and QRZ (pijul unrec XJY QRZ), recording again and pushing
• or, alternatively, pijul unrec XJY, and then pijul record --amend XJY.

does that seem about right? still very new with these workflows!

Perfect! I’m testing it right now. We’re actually all new to these workflows ;-)

It complains about settings, which should be setting, actually.

apologies. have amended the change

Thanks! I believe we can close this, feel free to reopen if not (people can still comment and push changes after the discussion is closed, it just doesn’t appear in the list anymore).