Page MenuHomePhabricator

[lib] Add logic for handling faonut actions
ClosedPublic

Authored by inka on Sep 15 2023, 7:04 AM.
Tags
None
Referenced Files
F3390093: D9217.id.diff
Fri, Nov 29, 10:08 PM
Unknown Object (File)
Thu, Nov 28, 2:26 AM
Unknown Object (File)
Wed, Nov 27, 1:38 PM
Unknown Object (File)
Mon, Nov 25, 6:24 PM
Unknown Object (File)
Fri, Nov 8, 8:52 AM
Unknown Object (File)
Fri, Nov 8, 6:36 AM
Unknown Object (File)
Oct 27 2024, 10:19 PM
Unknown Object (File)
Oct 27 2024, 10:19 PM
Subscribers

Details

Summary
Test Plan

Created an object for searchMessages:

const searchMessagesActionObj: KeyserverCall<
  [SearchMessagesRequest],
  SearchMessagesResponse,
> = {
  actionFunc: searchMessages,
  config: { keyserverSelection: 'fanout' },
};

Called this action from the message search modal, checked that the results appear, and that it is possible to load more results

Diff Detail

Repository
rCOMM Comm
Lint
No Lint Coverage
Unit
No Test Coverage

Event Timeline

inka requested review of this revision.Sep 15 2023, 7:21 AM

Overall looks good but I have a question. In the test plan you write:

const searchMessagesActionObj: KeyserverCall<
  [SearchMessagesRequest],
  SearchMessagesResponse,
> = {
  actionFunc: searchMessages,
  config: { keyserverSelection: 'fanout' },
};

but shouldn't the return type be [SearchMessagesResponse] as it's a fanout action?

@michal my idea is that every action will take care of the merging inside

lib/utils/keyserver-call.js
107 ↗(On Diff #31240)

I meant that this Return is wrong, because in case of fanout it returns [Return]

michal requested changes to this revision.Sep 26 2023, 3:20 AM

I just realized a problem with approach: we don't want to send all of the input to all of the keyservers. For example in D9240, in fetchSingleMostRecentMessagesFromThreads we want to send each keyserver only the ids that it "owns". Otherwise we could expose private date to other keyservers. We need some logic to split a single request into multiple request, one for each keyserver.

This revision now requires changes to proceed.Sep 26 2023, 3:20 AM
lib/utils/keyserver-call.js
97 ↗(On Diff #31606)

Should we replace Object with { +[string]: mixed }?

This revision is now accepted and ready to land.Oct 4 2023, 6:52 AM
lib/utils/keyserver-call.js
97 ↗(On Diff #31606)

In some cases, when an action used to take for example a string, I left the requests in the form of {[string]: string} to lessen the amount of boilerplate, so { +[keyserverID: string]: { +[string]: mixed } } won't work

ashoat requested changes to this revision.Nov 2 2023, 7:13 AM

Requesting changes to make sure I at least get a satisfactory explanation before this lands (ideally a proper solution)

lib/utils/keyserver-call.js
97 ↗(On Diff #31606)

Can you please clarify further why it "won't work"? Is the issue with { +[string]: mixed } that it's read-only? If so, you should see if you can convert the {[string]: string} cases to read-only. If that doesn't work, you can consider { [string]: mixed }.

I know it was Object before, but it's our responsibility as stewards of the codebase to always improve types when we touch them. Object is particularly bad, as it's basically synonymous with any. We should strive to remove all anys from the codebase.

This revision now requires changes to proceed.Nov 2 2023, 7:13 AM
lib/utils/keyserver-call.js
97 ↗(On Diff #31606)

I meant that if before the request was a string, then I left it as string, so the requests is in that case a { +[keyserverID]: string }, so not { +[keyserverID: string]: { +[string]: mixed } }

I can change this by either:

  • creating a type for every input, so an action that took string takes { +field: string }. Then the type of requests can be { +[keyserverID: string]: { +[string]: mixed } }
  • typing requests here as { +[keyserverID: string]: mixed }

The first option adds a lot of boilerplate, that's why I wanted to avoid it. The second option should probably have been used in the first place, since that's how CallKeyserverEndpoint is typed currently, and we have a task to update it at some point: ENG-4968

lib/utils/keyserver-call.js
97 ↗(On Diff #31606)

Thanks for explaining! Two follow-ups here:

  1. To get more context on option 1, can you list all of the cases where we pass something that isn't a POJO as data to callKeyserverEndpoint?
  2. Can you explain if anything is blocking option 2 at this point? The task you linked talks about improving FROM option 2, but it seems like option 2 itself should not be blocked on anything (but perhaps I'm missing something)
lib/utils/keyserver-call.js
97 ↗(On Diff #31606)

I went through all actions and actually it looks like I have been creating the auxiliary types. I got it mixed up with that I didn't want to create them for action inputs, not for requests. I'm very sorry about this, I will update the type

Update CallKeyserverEndpoint type

This revision is now accepted and ready to land.Nov 2 2023, 12:53 PM
This revision was automatically updated to reflect the committed changes.
lib/utils/keyserver-call.js
129–133 ↗(On Diff #32750)

Unclear why this promise stuff was necessary here... bindCallKeyserverEndpoint(keyserverID) does not return a Promise

lib/utils/keyserver-call.js
129–133 ↗(On Diff #32750)

Hmm actually, looking at this again I think I'm just confused by some naming

lib/utils/keyserver-call.js
129–133 ↗(On Diff #32750)

The only issue was actually mine in D9748... I typed promises as { [string]: Promise<CallServerEndpoint> }, when it is actually { [string]: Promise<Object> }.

I got confused by the naming here... bindCallKeyserverEndpoint seemed to imply that the function bound the CallKeyserverEndpoint function, but really it's making the server call to one specific keyserver.

Flow didn't catch it because Object is basically any