Order of port declarations in resulting JS influences behaviour

[Jtalk]

SSCCE

https://ellie-app.com/dfBrYLCyNPga1

Environment

• Elm 0.19.1
• Elm/Core 1.0.5

Trivia

• This seems to be a special case of #896
• We're sending an outgoing port command that synchronously responds via another (incoming) port.
• When an outgoing port command is issued together with a model change that triggers an incoming port subscription, the response to that port might get skipped if the outgoing port is registered first in the resulting JS.
• In the above example, the first message from the button gets lost because Subs of the receiving port get updated after the Cmd is fired.
• In the above example, renaming zsubscriptions to subscriptions puts it above update in the resulting JS, changing the behaviour of the button (Subs now come first & the incoming port's data is received correctly).

Problem

• The order of registration of incoming and outgoing ports in the resulting JS file influences behaviour for dynamic subscriptions.

Impact

• It makes synchronous data requests via Ports unpredictable as soon as we start making subscriptions dynamically.
• For me it breaks localStorage access from nested components (components request data on init and never get it back).

Possible fix

The problem seems related to how _Platform_dispatchEffects iterates over effect managers. It bundles Cmd and Sub updates into a single object & submits it to each manager in the order of their appearance in the managers object.

The fix would involve forcing the order, I see two options:

• Separate Sub and Cmd effects and submit them separately, iterating over the managers container twice (e.g. submit { __subs: subs, __cmds: __List_Nil } to each of the managers followed by { __subs: __List_Nil, __cmds: cmds }. The downside is that each manager would receive 2 events per lifecycle tick.
• Introduce the idea of manager ordering & set priority for managers handling incoming ports over outgoing. It would require a new field in the managers, but the change should remain private within Kernel.Platform.

Since the existing behaviour is unpredictable, I guess it can be seen as backward-compatible patch-level fix.

I'm happy to raise a PR if you think it's worth fixing

[github-actions]

Thanks for reporting this! To set expectations:

• Issues are reviewed in batches, so it can take some time to get a response.
• Ask questions a community forum. You will get an answer quicker that way!
• If you experience something similar, open a new issue. We like duplicates.

Finally, please be patient with the core team. They are trying their best with limited resources.