Make it compatible with Adapt 4 and Metal 1

[joelreymont]

Prevent RL from downgrading Metal 1 on the Mac, enabling Lux to use the Mac GPU.

[joelreymont]

Here's why I updated the umbrella package and undertook updating RL in the first place.

Lux was at 0.5.15 and Metal at 1.0 but the outdated umbrella package versions downgraded both Lux and Metal. More importantly, Lux does not use the GPU (Metal) on the Mac unless Metal is at 1.0.

Is there a way for me to use RL and my Mac GPU with Lux while keeping old RL sub-packages in the umbrella RL package?

What needs to be done to ship the umbrella RL with the latest packages?

(@v1.10) pkg> st
Status `~/.julia/environments/v1.10/Project.toml`
⌃ [b2108857] Lux v0.5.15
⌃ [dde4c033] Metal v0.5.0
  [158674fc] ReinforcementLearning v0.11.0-dev `~/Work/Julia/ReinforcementLearning.jl`
⌃ [02a925ec] cuDNN v1.1.1
Info Packages marked with ⌃ have new versions available and may be upgradable.

(@v1.10) pkg> add Metal@1
   Resolving package versions...
ERROR: Unsatisfiable requirements detected for package Adapt [79e6a3ab]:
 Adapt [79e6a3ab] log:
 ├─possible versions are: 0.3.0-4.0.1 or uninstalled
 ├─restricted by compatibility requirements with Metal [dde4c033] to versions: 4.0.0-4.0.1
 │ └─Metal [dde4c033] log:
 │   ├─possible versions are: 0.0.1-1.0.0 or uninstalled
 │   └─restricted to versions 1 by an explicit requirement, leaving only versions: 1.0.0
 └─restricted by compatibility requirements with ReinforcementLearningCore [de1b191a] to versions: 3.0.0-3.7.2 — no versions left
   └─ReinforcementLearningCore [de1b191a] log:
     ├─possible versions are: 0.1.0-0.13.0 or uninstalled
     └─restricted to versions 0.9 by ReinforcementLearning [158674fc], leaving only versions: 0.9.0-0.9.4
       └─ReinforcementLearning [158674fc] log:
         ├─possible versions are: 0.11.0 or uninstalled
         └─ReinforcementLearning [158674fc] is fixed to version 0.11.0-dev

[jeremiahpslewis]

Let's take this one at a time, first update the sub packages to the latest Adapt.jl, etc., which can be done immediately, then we can start talking about getting RL.jl in sync? To do this, please drop the changes as discussed in my review.

As a side note, are you actively using RL.jl / able to put time into the project? If so, contributions are very welcome and I can guide you through the most important chunks of work...

[joelreymont]

Everything works fine if the umbrella package is updated. Meaning that RL compiles and does not downgrade metal. I will re-test and I dropped the changes you requested. I plan to actively use RL and can put time into it. I’m new to Julia but I’m learning quickly.

…

On Thu, 29 Feb 2024 at 12:29, Jeremiah ***@***.***> wrote: Let's take this one at a time, first update the sub packages to the latest Adapt.jl, etc., which can be done immediately, then we can start talking about getting RL.jl in sync? To do this, please drop the changes as discussed in my review. As a side note, are you actively using RL.jl / able to put time into the project? If so, contributions are very welcome and I can guide you through the most important chunks of work... — Reply to this email directly, view it on GitHub <#1008 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAAESZ3FLBPYRPVQL5FIGULYV4BI7AVCNFSM6AAAAABD6HJYC2VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTSNZQHAZTSMBVGI> . You are receiving this because you authored the thread.Message ID: <JuliaReinforcementLearning/ReinforcementLearning.jl/pull/1008/c1970839052 @github.com>

[joelreymont]

I’m trying to use Julia and RL for drone AI. I have the time, let me know what needs to be done.

…

On Thu, 29 Feb 2024 at 12:58, Joel Reymont ***@***.***> wrote: Everything works fine if the umbrella package is updated. Meaning that RL compiles and does not downgrade metal. I will re-test and I dropped the changes you requested. I plan to actively use RL and can put time into it. I’m new to Julia but I’m learning quickly. On Thu, 29 Feb 2024 at 12:29, Jeremiah ***@***.***> wrote: > Let's take this one at a time, first update the sub packages to the > latest Adapt.jl, etc., which can be done immediately, then we can start > talking about getting RL.jl in sync? To do this, please drop the changes as > discussed in my review. > > As a side note, are you actively using RL.jl / able to put time into the > project? If so, contributions are very welcome and I can guide you through > the most important chunks of work... > > — > Reply to this email directly, view it on GitHub > <#1008 (comment)>, > or unsubscribe > <https://github.com/notifications/unsubscribe-auth/AAAESZ3FLBPYRPVQL5FIGULYV4BI7AVCNFSM6AAAAABD6HJYC2VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTSNZQHAZTSMBVGI> > . > You are receiving this because you authored the thread.Message ID: > <JuliaReinforcementLearning/ReinforcementLearning. > ***@***.***> >

[jeremiahpslewis]

As a first step, can you get the CI to pass?

Side note: for the time being, it's best to install the sub packages separately and NOT include the umbrella package RL.jl in your environment. This is also how the GitHub Actions CI is set up.

[joelreymont]

Let me take a look…

…

On Thu, 29 Feb 2024 at 13:50, Jeremiah ***@***.***> wrote: As a first step, can you get the CI to pass? Side note: for the time being, it's best to install the sub packages separately and NOT include the umbrella package RL.jl in your environment. This is also how the GitHub Actions CI is set up. — Reply to this email directly, view it on GitHub <#1008 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAAESZYEGETK5GTKGUQLQLTYV4KW5AVCNFSM6AAAAABD6HJYC2VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTSNZQHE3TCOJVGA> . You are receiving this because you authored the thread.Message ID: <JuliaReinforcementLearning/ReinforcementLearning.jl/pull/1008/c1970971950 @github.com>

[joelreymont]

Some tests are failing and OpenSpiel dependency is failing to compile.

I'm on it.

[jeremiahpslewis]

ok, a note, I need to tag and release new versions of RLTrajectories and CircularArrays before they are picked up here.

[joelreymont]

ok, a note, I need to tag and release new versions of RLTrajectories and CircularArrays before they are picked up here.

How does this affect me?

[jeremiahpslewis]

I think this is necessary...

• [OpenSpiel] rebuild with no changes for latest libcxxwrap JuliaPackaging/Yggdrasil#8219
• New version: OpenSpiel v0.1.5 JuliaRegistries/General#101984 (comment)

[joelreymont]

Issue #1010 prevents CI from passing.

[jeremiahpslewis]

this will fix it, need to wait for GitHub JuliaReinforcementLearning/OpenSpiel.jl#19

[jeremiahpslewis]

once this merges, rlenv is hopefully good to go: JuliaRegistries/General#102066

[joelreymont]

What's the next step?

[jeremiahpslewis]

I’ll release the new versions of RLCore / RLEnv. Can you track down the bug in the offlineagents?

[joelreymont]

I'm on it.

…

On Fri, Mar 1, 2024 at 6:26 PM Jeremiah ***@***.***> wrote: I’ll release the new versions of RLCore / RLEnv. Can you track down the bug in the offlineagents? — Reply to this email directly, view it on GitHub <#1008 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAAESZ4RVO4FUF6MVDNRLUTYWCT37AVCNFSM6AAAAABD6HJYC2VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTSNZTGQ4TANRWHE> . You are receiving this because you authored the thread.Message ID: <JuliaReinforcementLearning/ReinforcementLearning.jl/pull/1008/c1973490669 @github.com>