[JSTC-37] Create table help output shows prefix

[sanderpick]

When creating a table, I noticed that the prefix is auto-applied from the table name in the create statement (neat!). However, the help command still shows that you can supply a prefix. I'm not sure if that still actually works. In either case, I don't think there should be two ways to do this… or if that is needed for some reason, the help command should be clear about it.

tableland create [schema] [prefix]        Create a new table

cc @carsonfarmer @joewagner

_{From SyncLinear.com | JSTC-37}

[joewagner]

@sanderpick good call on this command being unclear. I'm dumping my thoughts below...

For context, I believe we have 5 ways to run the create command:

1. provide a schema and prefix via command args
2. provide a schema via command args. (note: table name will not have a prefix, e.g. _31337_3. This feature was added by a request that may have originated in discord. Details: [JSTC-27] Don't require prefix flag for create command with schema js-tableland-cli#306)
3. provide a full create query. (note: this can be any number of create statements, i.e. you can create all your tables with one command)
4. call the command with no arguments, which opens a stdin prompt allowing you to freely type. (note: I just tried this out and I'm not sure this works right. If we want to keep this feature I can open an issue to fix)
5. use the command to run a file that contains SQL.

The full text of npx tableland create --help tries to explain to explain all of this, but maybe still confusing.

Ideas:

• Reduce the number of ways to use create.
• Replace the current usage with a different single usage.
• Make changes to the help messages.
• other?
  cc: @dtbuchholz @awmuncy

[carsonfarmer]

I think input from stdin/files is still useful. We should just drop support for extra flags etc though (IMO). Because there should really just be one way to do it, using a create statement. Stdin/files are also just doing it the same way (via a create statement), they are just alternative input sources.

[carsonfarmer]

I think the stdin option requires you to pipe in the content? But not 100% sure anymore?

[joewagner]

I think the stdin option requires you to pipe in the content? But not 100% sure anymore?

That makes sense. You can just type, but the prompt never closes and only works for the first statement.

[carsonfarmer]

Does Ctrl+D properly close it out?

[joewagner]

Does Ctrl+D properly close it out?

Yeah, which is all good, it's just a little confusing since the second (and third etc) command don't do anything.

[joewagner]

I think input from stdin/files is still useful. We should just drop support for extra flags etc though (IMO). Because there should really just be one way to do it, using a create statement. Stdin/files are also just doing it the same way (via a create statement), they are just alternative input sources.

Sounds good to me, but this would undo the changes that fixed tablelandnetwork/js-tableland-cli#306 so maybe @dtbuchholz has thoughts.

[dtbuchholz]

I think that should be fine...I should be able to pass an empty prefix within a full create statement anyways like create table "" ...

[sanderpick]

Ah, makes sense! I didn't even realize that i had supplied a full create statement. That is a nice pattern. The other way that makes sense to me is with the shell command => just type SQL as normal, including create statements. That's possible, right?

I don't have a strong pref on removing options here... but we should document them. If I do tableland create --help I just see:

tableland create -h
tableland create [schema] [prefix]

Create a new table

Positionals:
  schema  SQL table schema, or full create statement (skip to read from stdin)
                                                                        [string]

Options:
  -h, --help                 Show help                                 [boolean]
  -V, --version              Show version number                       [boolean]
      --baseUrl              The URL of your Tableland validator        [string]
  -c, --chain                The EVM chain to target                    [string]
      --enableEnsExperiment  Enable ENS experiment                     [boolean]
      --ensProviderUrl       Enable ENS experiment                      [string]
  -k, --privateKey           Private key string                         [string]
  -p, --providerUrl          JSON RPC API provider URL. (e.g., https://eth-rinke
                             by.alchemyapi.io/v2/123abc123a...)         [string]
      --prefix               Table name prefix (ignored if full create statement
                              is provided)                              [string]
      --ns                   ENS namespace to resolve schema from (experimental)
                                                                        [string]
  -f, --file                 Get statement from input file

I think some examples of the different patterns would be nice in the help output.

[dtbuchholz]

(moved to new issue #90)