@greg-szabo:
The confusion comes from the two commands with different purpose and similar spelling:
gaiad init gentx
gaiad init --gen-txs
so I don't bother remembering which command does what...
After talking through this with @ValarDragon it sounds like the we might want to move the --gen-txs to a gaiad init gen-genesis
It's up to you guys, but from an ops perspective, the gen-tx command shouldn't be a subcommand of the init command. Something like
gaiad gen-tx to create an initial transaction
gaiad init --with-tx for a generic set of config files that include the generated initial transactions...
I'm into @greg-szabo 's proposal
It seems like the case of generating the genesis file is a bit error prone and there is manual editing that happens frequently. If users all did a gaiad init --with-tx to gen their genesis I would say yes, but because the genesis is frequently distributed it makes more sense in my head for this to be its own command instead of hiding behind init. Those were my thoughts.
The --with-tx parameter (instead of a subcommand) only makes sense, because it's already the init command's responsibility to generate a genesis.json file, together with config.toml. Logically it makes sense that the plain command creates a generic set for a local testnet, while the parameter is a modifier to create the same set with different rules.
In any case, this is more for people generating their own testnet config, rather than for the public testnet or mainnet validators. As @jackzampolin mentioned, in the gaia-* testnet and the mainnet case, the genesis file is manually crafted. The only important part for the validator is to be able to generate an initial transaction. (Hence I thought it makes more sense to make it its own command.)
As I said, it's up to you all.
I like what @greg-szabo suggested in regards to gaiad gen-tx and gaiad init --with-txs. Nomenclature could be up for discussion but I think it's fine and lends itself to being more intuitive...it took me a while to learn I could even generate a genesis file from a bunch of txs btw.
Yeah I agree with @alexanderbez there. That functionality was not clear to me initially. Another benefit of having the gen-genesis command under init would be that we could have help text in the CLI for it.
At that point, why not just have gaiad init genesis?
I'm good with that. Any objections?
I think we should go ahead with gaiad init genesis. I think we can further rename commands #postlaunch.
gaiad init will also generate a genesis.json. It is misleading.
Hmmm, I see. Indeed misleading. Then I suppose it should just be a flag as you mentioned, huh?
I don't think theres a good answer here. What is obvious is that the current situation is untenable. My personal inclination would be to go for gaiad init genesis. Lets vote on that proposal 馃憤 for yes and 馃憥 for no.
Most helpful comment
I don't think theres a good answer here. What is obvious is that the current situation is untenable. My personal inclination would be to go for
gaiad init genesis. Lets vote on that proposal 馃憤 for yes and 馃憥 for no.