Cell Info design for multi instance use on the client

[nphias]

export type InstalledCell = { cell_id: CellId; role_name: RoleName; };

export type InstalledCell = { cell_id: CellId; cell_name: CellName; role_name: RoleName; };

the yaml files appear to have "name" and "role" fields which have two different intentions
If i have two cell clones it would be nice to have a name field which is common (ie the cell_name) and another which has meaning for that instance assuming "role" could be used. One can then display a list of cells of the same type in the UI distinguisable by that instance field

[jost-s]

InstalledCell is being replaced by CellInfo. To be released this coming week.

[jost-s]

This is going to be made available in the client tomorrow.

[jost-s]

These are the exact fields that will be returned for clone cells: https://github.com/holochain/holochain/blob/b60c64ec375f88670a53af11edd160047f37eff7/crates/holochain_conductor_api/src/app_interface.rs#L257-L272

[nphias]

/// Provisioned cell, a cell instantiated from a DNA on app installation.
#[derive(Clone, Debug, Eq, PartialEq, serde::Serialize, serde::Deserialize)]
pub struct ProvisionedCell {
    pub cell_id: CellId,
    pub dna_modifiers: DnaModifiers,
    pub name: String,
}

/// Cloned cell that was created from a provisioned cell at runtime.
#[derive(Clone, Debug, Eq, PartialEq, serde::Serialize, serde::Deserialize)]
pub struct ClonedCell {
    pub cell_id: CellId,
    pub clone_id: CloneId,
    pub original_dna_hash: DnaHash,
    pub dna_modifiers: DnaModifiers,
    pub name: String,
    pub enabled: bool,
}

ok i see in the docs clone_id maps to role.. what if i want to run two instances of a provisioned cell, one as a user and the other as an admin (different agent keys)? would that mean making a clone first?

Why not make a more consistent structure, so on the front end devs can create stores and receptor services per cell instance without having to worry about the cell type? ie [cell_name:cell_instance]

ProvisionedCell {
    pub cell_id: CellId,
    pub dna_modifiers: DnaModifiers,
    pub name: String,
    pub instance: String,
}

ClonedCell {
    pub cell_id: CellId,
    pub original_dna_hash: DnaHash,
    pub dna_modifiers: DnaModifiers,
    pub name: String,
    pub instance: String,
    pub enabled: bool,
}

[nphias]

another issue / feature request .. which i think makes sense to add here, Is that it would be nice to have the cell_api as a cellInfo property.. then devs can programatically infer that the cells functions ( RECEPTOR API ) match those functions in their Receptor service classes . Receptors span multiple zomes and are congruent with a design centered on a Cell based biomemetic architecture

[jost-s]

@maackle @abe-njama FYI, up for discussion (where?)

[maackle]

I just saw that GH auto-closed this issue in my name but I have no idea how that happened. I'll reopen since there is still potential for discussion.

[mattyg]

another issue / feature request .. which i think makes sense to add here, Is that it would be nice to have the cell_api as a cellInfo property.. then devs can programatically infer that the cells functions ( RECEPTOR API ) match those functions in their Receptor service classes . Receptors span multiple zomes and are congruent with a design centered on a Cell based biomemetic architecture

I created an issue to track this in the holochain repo holochain/holochain#4453 -- I believe it captures the intent of this request but feel free to add any more context or clarification there - or let me know if I'm way off base. @nphias

ok i see in the docs clone_id maps to role.. what if i want to run two instances of a provisioned cell, one as a user and the other as an admin (different agent keys)? would that mean making a clone first?

Yes if you want to have two cell instances of the same DNA, you could clone the provisioned cell. Alternatively I believe you could include the DNA twice in your app manifest with different role names and properties. Does that fulfill your requirements? @nphias

Why not make a more consistent structure, so on the front end devs can create stores and receptor services per cell instance without having to worry about the cell type? ie [cell_name:cell_instance]

ProvisionedCell {
   pub cell_id: CellId,
   pub dna_modifiers: DnaModifiers,
   pub name: String,
   pub instance: String,
}

ClonedCell {
   pub cell_id: CellId,
   pub original_dna_hash: DnaHash,
   pub dna_modifiers: DnaModifiers,
   pub name: String,
   pub instance: String,
   pub enabled: bool,
}

I'm not sure I understand this. Is this underlying issue that it is not ergonomic to match cells by the role of their source cell?

Would this be addressed by having a consistent identifier for all clones that were cloned from the same cell? For example, by including the original role name in all cells, and then including the clone_index as a separate field just in clones? If so I believe its covered by holochain/holochain#3219 -- does that make sense @nphias or is this a distinct issue?