Skip to content

fix: revert differences in ModuleEnvironment from v5#112

Merged
jdrueckert merged 1 commit intorelease/v7.xfrom
v7/spike/modenv-vs-v5
Jun 7, 2021
Merged

fix: revert differences in ModuleEnvironment from v5#112
jdrueckert merged 1 commit intorelease/v7.xfrom
v7/spike/modenv-vs-v5

Conversation

@keturn
Copy link
Member

@keturn keturn commented Apr 30, 2021

There were a few changes in how ModuleEnvironment builds reflections and permissions between v5 and v7.

They are the sort of thing that look like they probably do make sense in many situations, but I think Terasology has things that depend on the old behavior.

I doubt we should really revert these entirely, but what should we do?

@keturn keturn added Type: Question Issue intended to help understanding something that is unclear Topic: Module Requests, Issues and Changes related to gestalt modules v7 labels Apr 30, 2021
@keturn
Copy link
Member Author

keturn commented May 1, 2021

@keturn
Copy link
Member Author

keturn commented May 8, 2021

Notes from today's meeting:

  • The only stakeholder in gestalt-module other than Terasology is Destination Sol.
    • there aren't really other people we can talk to about what the impact of a change like this would be
    • but that also means there aren't other things to worry about breaking
  • confusion about what exactly this change broke / what would be fixed if we revert it
  • keturn thinks we can work around the change to createPermissionProviderFor at the application level, but
  • the change within buildFullReflections can't be overridden at another level
    • if necessary, we'll revert it, and put a note in some Architecture Decisions Log (which lives where?)

@keturn
Copy link
Member Author

keturn commented May 8, 2021

I guess the other option besides reverting it would be to twiddle some things around so the application can customize that behavior in a ModuleEnvironment subclass. But there's not much sense in doing that if we can't articulate why an application would choose one way over the other.

@keturn keturn force-pushed the v7/spike/modenv-vs-v5 branch from b8a593c to 29ec08f Compare June 5, 2021 18:18
@keturn keturn changed the title differences in ModuleEnvironment from v5 fix: revert differences in ModuleEnvironment from v5 Jun 5, 2021
@keturn keturn marked this pull request as ready for review June 5, 2021 18:19
@jdrueckert
Copy link
Member

jdrueckert commented Jun 6, 2021

Tested the following things before (release/v7.x) and after (v7/spike/modenv-vs-v5) this change:

  • CoreGameplay -> looks the same both before and after
  • Malicious tests -> before 1/2, after 2/2
  • integration tests -> Health fails both before and after, new failure after in ModuleTestingEnvironment ModuleTestingEnvironment also fails before and after

I'm not really sure how to test a dist zip of iota. I tried with ./gradlew distZip but that gives me a .zip with the engine only.
Also I don't know whether there's further things we should test for this...?

FYI @skaldarnar @keturn

@jdrueckert
Copy link
Member

Pre-GSoC Release

@jdrueckert
Copy link
Member

To test a dist zip of iota I ended up doing the following:

  1. ./gradlew distZip
  2. copy facades/PC/build/distributions/Terasology-5.0.0-SNAPSHOT.zip some place else and unzip
  3. copy module jars (from <module>/build/libs/) into <unzip dir>/modules/
  4. start Terasology and start CoreGameplay

I observed that the game starts fine and looks as expected

@jdrueckert jdrueckert merged commit 118a29f into release/v7.x Jun 7, 2021
@jdrueckert jdrueckert deleted the v7/spike/modenv-vs-v5 branch June 7, 2021 21:46
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

Topic: Module Requests, Issues and Changes related to gestalt modules Type: Question Issue intended to help understanding something that is unclear

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants