Page MenuHomePhabricator

[native] validateAndGetPrekeys method
ClosedPublic

Authored by varun on Feb 1 2024, 9:45 PM.
Tags
None
Referenced Files
Unknown Object (File)
Sat, Nov 23, 2:29 AM
Unknown Object (File)
Sat, Nov 23, 1:24 AM
Unknown Object (File)
Sat, Nov 16, 12:15 PM
Unknown Object (File)
Sat, Nov 16, 10:20 AM
Unknown Object (File)
Thu, Nov 14, 8:14 AM
Unknown Object (File)
Thu, Nov 14, 8:03 AM
Unknown Object (File)
Wed, Nov 13, 7:37 PM
Unknown Object (File)
Wed, Nov 13, 6:52 PM
Subscribers

Details

Summary

New method to validate and get content and notif prekeys.

Test Plan

called the method multiple times from js and confirmed the prekey values did not change

Diff Detail

Repository
rCOMM Comm
Lint
No Lint Coverage
Unit
No Test Coverage

Event Timeline

varun requested review of this revision.Feb 1 2024, 10:04 PM

clean up the C++ code a little

Leaving only comments to let others review as well

native/cpp/CommonCpp/NativeModules/CommCoreModule.cpp
665 ↗(On Diff #36568)

when we have only two elements I think it's better to use std::pair:

  • std::tuple is built to handle more elements which means could be slower/worse in terms of memory used
  • using member values first and second seems more straightforward than calling methods
  • when see a pair, I immediately know that there are exactly two items involved, with std::tuple it could be any number

But I feel like like those are more personal preferences so if you have a different opinion std::tuple also makes sense to me

794 ↗(On Diff #36568)

in this case, you have to also call this->persistCryptoModule();

native/schema/CommCoreModuleSchema.js
88 ↗(On Diff #36568)

I think we could deprecate generateAndGetPrekeys to avoid misuse in the future

use std::tie in existing method

varun planned changes to this revision.Feb 2 2024, 8:47 AM
varun added inline comments.
native/cpp/CommonCpp/NativeModules/CommCoreModule.cpp
665 ↗(On Diff #36568)

i can refactor to use pair, i don't have a strong preference

794 ↗(On Diff #36568)

thanks, didn't realize this modifies the account

native/schema/CommCoreModuleSchema.js
88 ↗(On Diff #36568)

sure i'll introduce a diff after this to deprecate

I think we could deprecate generateAndGetPrekeys to avoid misuse in the future

Yeah, it looks like its only callsite can be replaced with the new validateAndUploadPrekeys.

It also looks like CryptoModule::generateAndGetPrekey can be deprecated.

native/cpp/CommonCpp/NativeModules/CommCoreModule.cpp
780 ↗(On Diff #36595)

If this the right condition? In validateAndUploadPrekeys, we appear to check has_value.

791 ↗(On Diff #36595)

Nit: I would add another newline here, since the contentPrekeySignature is more associated with the contentPrekey than the other two variables

It also looks like CryptoModule::generateAndGetPrekey can be deprecated.

We actually can't deprecate it since we use it in CryptoModule::validatePrekey, but we can move it out of the public interface. We also don't need NotificationsCryptoModule::generateAndGetNotificationsPrekey anymore

native/cpp/CommonCpp/NativeModules/CommCoreModule.cpp
780 ↗(On Diff #36595)

Yes, according to cppreference:

When an object of type optional<T> is contextually converted to bool, the conversion returns true if the object contains a value and false if it does not contain a value.

This also explicitly says that explicit operator bool() const and bool has_value() const are synonymous: https://en.cppreference.com/w/cpp/utility/optional/operator_bool

addressing feedback. will handle code deprecation in the next diff

ashoat requested changes to this revision.Feb 3 2024, 9:35 AM
ashoat added inline comments.
native/cpp/CommonCpp/NativeModules/CommCoreModule.cpp
692 ↗(On Diff #36601)

I think this code is wrong for the below reason too. @marcin, can you confirm?

786 ↗(On Diff #36601)

I think we need to persistCryptoModule() in this case too. Even if validatePrekey() returns false, it can still call forgetOldPrekey(), so we always need to call persistCryptoModule() after calling validatePrekey()

Maybe we should just extract the persistCryptoModule calls into a single one on line 791?

This revision now requires changes to proceed.Feb 3 2024, 9:35 AM
native/cpp/CommonCpp/NativeModules/CommCoreModule.cpp
692 ↗(On Diff #36601)

I think you ar right. We should call persistCryptoAccount() after validation without any condition.

call persistCryptoModule unconditionally since validatePrekey can call forgetOldPrekey, which modifies the olm account

The code is technically correct but I think it needs some adjustments so that it conforms to our C++ conventions.

native/cpp/CommonCpp/NativeModules/CommCoreModule.cpp
703

I've just realized that there is another bug in this function. Since you've already modified this function, could you put those lines here?

if(this->cryptoModule == nullptr) {
    this->jsInvoker_->invokeAsync([=, &innerRt]() {
        promise->reject("user has not been initialized");
    });
    return;
}

Basically in this function we don't check if cryptoModule has been initialized and we should early reject and return in case it hasn't been.

788

There is too much indentation here. In this line I think you should reject the promise and return early. Then you could remove else wrapping around the try ... catch.

792

Technically correct, but we don't use this style of if statements in the codebase. I think you should just assign contentPrekey = this->cryptoModule->validatePrekey() right away and then conditionally perform other assignments if contentPrekey doesn't have value.

native/cpp/CommonCpp/NativeModules/CommCoreModule.cpp
703

This task should be prioritized to avoid such mistakes in future.

Agree with @marcin's comments about C++ conventions. Please make sure to adjust those before landing, and to add the this->cryptoModule == nullptr check he requested as well

This revision is now accepted and ready to land.Feb 7 2024, 10:07 AM
varun marked 4 inline comments as done.Feb 7 2024, 9:22 PM
This revision was automatically updated to reflect the committed changes.