Open Hub
Open Hub

 Settings |  Report Duplicate

1
I Use This!
Activity Not Available

Commits : Listings

Analyzed about 1 year ago. based on code collected about 1 year ago.
Jan 16, 2023 — Jan 16, 2024
Commit Message Contributor Files Modified Lines Added Lines Removed Code Location Date
Messages are no longer encoded as raw byte arrays.
akwizgran
More... over 13 years ago
Defined some user-defined tags for protocol elements. Currently they're just adding redundancy, but in future they'll be used for parsing nested elements.
akwizgran
More... over 13 years ago
Read and write user-defined tags.
akwizgran
More... over 13 years ago
Fun fact: it's never worth writing a length as an int8.
akwizgran
More... over 13 years ago
A more efficient encoding for short strings, raws, lists and maps. Now we can encode a list of three small integers in 4 bytes like MessagePack does, should that ever turn out to be useful.
akwizgran
More... over 13 years ago
Cleaned up serial and protocol packages in preparation for user-defined types.
akwizgran
More... over 13 years ago
Check that EOF occurs when expected.
akwizgran
More... over 13 years ago
Added DroidSansFallback font for platforms without CJK fonts installed.
akwizgran
More... over 13 years ago
Changed Eclipse project name to match repo name.
akwizgran
More... over 13 years ago
Only check the beginnings of font names, since the exact names vary by platform.
akwizgran
More... over 13 years ago
Upgraded to JUnit 4.9b3 for a Java 1.5 compatibility bugfix.
akwizgran
More... over 13 years ago
Don't use FileUtils.getBriarDirectory() in tests.
akwizgran
More... over 13 years ago
Don't use the real Briar directory for unit tests.
akwizgran
More... over 13 years ago
Updated the H2 jar to a version that hopefully works on OSX 10.4 (since it was built there).
akwizgran
More... over 13 years ago
Use commons-io to measure free space.
akwizgran
More... over 13 years ago
IO, IO, it's off to work we go.
akwizgran
More... over 13 years ago
Java 1.5 compatibility changes (for OSX 10.4). We depend on commons-io for a single method... might be worth copying the source into FileUtils if the license permits.
akwizgran
More... over 13 years ago
Don't store subscription or transport updates that are older than those already received. Also some small changes to DatabaseComponent impls for readability.
akwizgran
More... over 13 years ago
Retrieve messages from the database in raw form to avoid creating unnecessary short-lived objects. Added timestamps to headers.
akwizgran
More... over 13 years ago
Replaced clearSubscriptions() and addSubscription() with setSubscriptions().
akwizgran
More... over 13 years ago
New retransmission mechanism, which does away with the need for bundle IDs and should cope better with high bandwidth-delay product links.
akwizgran
More... over 13 years ago
Guice stuff.
akwizgran
More... over 13 years ago
Updated buildfile for unit tests.
akwizgran
More... over 13 years ago
Rewrote the bundle reading and writing code to eliminate copying. Signatures and digests are now calculated on the fly as the data is read or written. This is a little bit tricky in the case of reading because ReaderImpl uses a lookahead byte, so the signature and message digest need to lag one byte behind.
akwizgran
More... over 13 years ago
Batch and header builders should set the length to include the signature.
akwizgran
More... over 13 years ago
Added accessors for the amount of raw data read and written by readers and writers - this fixes a fixme in MessageParserImpl.
akwizgran
More... over 13 years ago
Message parser and encoder.
akwizgran
More... over 13 years ago
Check the return value from Signature.verify(). *cough*
akwizgran
More... over 13 years ago
Builders for incoming and outgoing headers and batches. The protocol and serial components can now be used to serialise, sign, deserialise and verify real bundles (except for message parsing).
akwizgran
More... over 13 years ago
Protocol refactoring. Each bundle now consists of a signed header and zero or more signed batches. There is no overall signature on the bundle, since the bundle's contents may need to be processed before the entire bundle has been read. The protocol does not prevent an adversary from removing batches from a bundle, reordering batches, moving them from one bundle to another, etc. However, since each batch is signed and acknowledged independently, no such guarantees are required. Bundle IDs will go away when the retransmission mechanism is changed.
akwizgran
More... over 13 years ago