Implement override triggers. #825
Conversation
…30.triggers-class
…gers.40.deep-sleep-state
…gers.40.deep-sleep-state
…iggers.50.deep-sleep-state2
src/service/containers.toit
Outdated
| @@ -240,6 +266,23 @@ class ContainerJob extends Job: | |||
| stringify -> string: | |||
| return "container:$name" | |||
|
|
|||
| deep-sleep-state -> any: | |||
| job-state := super | |||
There was a problem hiding this comment.
job-state <> deep-sleep-state
src/service/containers.toit
Outdated
| deep-sleep-state -> any: | ||
| job-state := super | ||
| if not override-triggers_: return job-state | ||
| return { |
There was a problem hiding this comment.
This looks clean, but we're only storing in RTC memory so we don't have to worry about upgradability. Maybe just stick to lists that are much shorter when encoded?
| @@ -24,3 +24,7 @@ class ContainerCurrent implements artemis.Container: | |||
| triggers -> List: | |||
| encoded-triggers := client_.container-current-triggers | |||
| return encoded-triggers.map: artemis.Trigger.decode it | |||
|
|
|||
| set-next-start-triggers triggers/List -> none: | |||
src/service/containers.toit
Outdated
| kind := trigger.kind | ||
| if kind == Trigger.KIND-BOOT: | ||
| trigger-boot = true | ||
| if kind == Trigger.KIND-INSTALL: |
There was a problem hiding this comment.
else if here and elsewhere.
src/service/containers.toit
Outdated
| @@ -219,6 +244,7 @@ class ContainerJob extends Job: | |||
| is-background_/bool := false | |||
|
|
|||
| triggers_/Triggers := Triggers | |||
| override_triggers_/Triggers? := null | |||
There was a problem hiding this comment.
I'd consider calling these triggers-override_, but if nothing else I'd kebabify.
src/service/containers.toit
Outdated
| return triggers_.to-encoded-list | ||
| return active-triggers.to-encoded-list | ||
|
|
||
| active-triggers -> Triggers: |
There was a problem hiding this comment.
Active sounds a little bit like they've fired already. Maybe effective? Also, I'd consider just having a field with the effective triggers and then have the default ones hide behind that.
triggers_/Triggers
triggers-default_/Triggers := Triggers
Fewer null checks and helper methods.
There was a problem hiding this comment.
Changed to two fields, where triggers-armed_ is usually identical to the triggers-default_ but may be overridden.
| If $triggers is set to null, then the original triggers are | ||
| restored. | ||
| */ | ||
| set-next-start-triggers triggers/List? -> none |
There was a problem hiding this comment.
Not sure yet, but I feel like I want to couple this with the container restart mechanism.
artemis.Container.current.restart --triggers=[ ... ] --delay=(Duration --s=10)
There was a problem hiding this comment.
We can add this functionality, but Consibio wants to be able to disable a pin-trigger. They might not necessarily need to restart at that point.
It's also easier if the program decides at start-up what its next triggers should be. (Or maybe just after connecting...).
If we force a restart when setting the trigger, then it's quite easy to forget it, or to accidentally not set if if the program crashes...
src/service/service.toit
Outdated
| @@ -155,6 +165,9 @@ class ArtemisServiceProvider extends ChannelServiceProvider | |||
| container-current-triggers -> List: | |||
| unreachable // Here to satisfy the checker. | |||
|
|
|||
| container-current-set-triggers new-triggers/List -> none: | |||
| unreachable // Here to satisfy the checker. | |||
There was a problem hiding this comment.
| unreachable // Here to satisfy the checker. | |
| unreachable // Here to satisfy the checker. |
src/service/service.toit
Outdated
| @@ -155,6 +165,9 @@ class ArtemisServiceProvider extends ChannelServiceProvider | |||
| container-current-triggers -> List: | |||
| unreachable // Here to satisfy the checker. | |||
There was a problem hiding this comment.
| unreachable // Here to satisfy the checker. | |
| unreachable // Here to satisfy the checker. |
No description provided.