[x/programs] refactor init #432
Conversation
| @@ -8,6 +8,8 @@ import ( | |||
| ) | |||
|
|
|||
| type Runtime interface { | |||
| // Create initializes and calls the init function. | |||
| Create(context.Context, []byte) (uint64, error) | |||
There was a problem hiding this comment.
I don't believe this is actually a necessary part of the runtime interface. Once the runtime is initialized then the user can make whatever call they want. My understanding is that a programID would be the result of a transaction generated by the VM.
| @@ -44,6 +44,27 @@ type runtime struct { | |||
| log logging.Logger | |||
| } | |||
|
|
|||
| func (r *runtime) Create(ctx context.Context, programBytes []byte) (uint64, error) { | |||
| err := r.Initialize(ctx, programBytes) | |||
There was a problem hiding this comment.
why isn't this part of Initialize Runtime?
| programID := initProgramStorage() | ||
|
|
||
| // call initialize | ||
| result, err := r.Call(ctx, "init", uint64(programID)) |
There was a problem hiding this comment.
init in this context is perhaps required by the examples as they are now but as stated above the programID is going to be the result something like a create_program action. I would like us moving towards that direction in each iteration if possible.
| // Initialize the program with no fields | ||
| Program::new().into() | ||
| true | ||
| } |
There was a problem hiding this comment.
I think the fact that we end up with a hard coded bool here speaks to the fact that init should not be a requirement.
Removes
initialize_programhost function and theProgramstruct in Rust.Instead we generate and initialize the programs storage and id in
goand inject it into theinitfunction.Changes
Programfolderinitialize_functionCreatemethod to the go runtime that sets up global storage and callsiniton the programCreate"publishes" the program, whileInitializegrabs the published program.Context