Stateless enviroments

[AlexandraVolokhova]

There're several methods in the environments which work with states (e.g. state2policy, get_parents) and either take state as an input or use internal state of the enrolment self.state. This was a source of bugs in the past (when an env was supposed to transform input states but is was transforming the internal state or vice versa)

Proposal (by @carriepl ):
Alternative approach would be to get rid of the internal state of the environment and require input state for all methods where the state is needed.

[alexhernandezgarcia]

I think I would like to hear more arguments for this change. If we make the environment stateless, then we should also get rid of self.done and pass it to all methods too, and probably self.n_actions, etc. It should be all or nothing, no?

[AlexandraVolokhova]

@carriepl could you elaborate on that? I'm not sure I fully understood your suggestion to be able to address Alex's question

[josephdviviano]

I would second this change Joseph (Mobile)

…

On Wed, Jul 10, 2024 at 12:30 Alexandra ***@***.***> wrote: @carriepl <https://github.com/carriepl> could you elaborate on that? I'm not sure I fully understood your suggestion to be able to address Alex's question — Reply to this email directly, view it on GitHub <#334 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AA7TL2UO3C3GNAEBWGTTUPTZLVOQ5AVCNFSM6AAAAABKVE72ZGVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDEMRQHE3TEOJSGM> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[alexhernandezgarcia]

I'm not boldly against it, but it's a huge change that would require changing many methods in all environments and all their units tests. Even if I wanted to have this change done in theory, I don't see myself doing it right now.

[josephdviviano]

Yes it makes sense to deprioritize it relative to the effort involved (unless it’s possible to support both simultaneously) Joseph (Mobile)

…

On Wed, Jul 10, 2024 at 13:29 Alex ***@***.***> wrote: I'm not boldly against it, but it's a huge change that would require changing many methods in *all* environments and all their units tests. Even if I wanted to have this change done in theory, I don't see myself doing it right now. — Reply to this email directly, view it on GitHub <#334 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AA7TL2RCIVPY2VULQFYBUEDZLVVQJAVCNFSM6AAAAABKVE72ZGVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDEMRRGA3TSNZYHE> . You are receiving this because you commented.Message ID: ***@***.***>