Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Patch v0.62.0 #20

Merged
merged 333 commits into from
Nov 22, 2024
Merged

Patch v0.62.0 #20

merged 333 commits into from
Nov 22, 2024

Conversation

tiainen
Copy link

@tiainen tiainen commented Nov 22, 2024

No description provided.

jrose-signal and others added 30 commits August 14, 2024 09:16
Add a wrapper around a block-based transport that encrypts and decrypts using 
an established Noise connection. Add a handshaker that establishes an encrypted 
Noise connection using the server-defined protocol. Implement the block-based 
transport trait on top of a websocket connection.

Co-authored-by: Jordan Rose <jrose@signal.org>
Even in Validate mode, we still need real maps to resolve IDs. The
HashMap/HashSet flexibility has moved to the *Data associated types of
the ReferencedTypes trait.
Since the usual constructor is strongly typed, this can only happen
when the serialized data is corrupted. But that is a possibility.
Convert the *Key and keyPair properties on *PreKeyRecord to throwing
methods.

Since the usual constructor is strongly typed, these should only throw
when the serialized data is corrupted. But that is a possibility.
Lots of JNI calls can throw Java exceptions, not just those that
directly invoke Java methods. By making jni's Error not directly
convert to our own error types, calling code is encouraged to check
for exceptions on any JNI call.
jrose-signal and others added 29 commits November 5, 2024 13:45
This will be off by default, unlike in previous releases of libsignal.
This saves potentially-costly calls to `read` across each bridge,
especially if the app language side *isn't* buffering. The downside is
that if the app side *is* buffering, we're spending more memory for
less gain. But that's on the order of kilobytes, not megabytes.
    ./gradlew backup-tool:run --args "--hmac-key abcd --aes-key cdef
      path/to/backup"
Ideally logs wouldn't be frequent enough for this to matter,
but if they are, we shouldn't pay the price.
We weren't passing it along to the app anyway.
Co-authored-by: Ruben De Smet <ruben.de.smet@rubdos.be>
If a client has an unusual timestamp for one message in a
conversation, it's likely to have many such messages. Dumping a
screenful of logs isn't going to help with that.
This gains us a few percent of performance across a wide variety of 
operations, including 5% off of sealed sender decrypt, 10% off of
backup validation, and double digits off of double ratchet decryption
of successive messages (inner ratchet only).

Unfortunately, the sha2-asm implementation crate doesn't support
Windows. Hopefully sha2 0.11 is released soon for us to update to.
Co-authored-by: Jordan Rose <jrose@signal.org>
- Several timestamp fields become `optional`
- Pin order also becomes `optional`, but 0 is still accepted for
  compatibility with existing tests
- Contact now has `identityKey` and `identityState`
- Quote type can be VIEW_ONCE
The x86 Android linker isn't happy with the object files, and that's
our only x86 platform, and it's just the emulator anyway.
@tiainen tiainen merged commit d58c28b into main Nov 22, 2024
2 of 12 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.