[dependabot]

Bumps github.com/xssnick/raptorq from 1.5.1 to 1.6.0.

Release notes

Sourced from github.com/xssnick/raptorq's releases.

v1.6.0

• About 30-35% faster decoding and encoder creation for 1MB and 16MB blocks, ~20% faster repair symbol generation
• Fix Decoder replacing the known-zero padding rows (K..K'-1) with symbols received above K, so reception overhead now lowers the decode failure probability as RFC 6330 expects
• Compute HDPC products in a single streaming pass with one accumulator row instead of scattering the input into a temporary matrix and chaining it in place
• Fold the HDPC-right product into the HDPC(E) pass by feeding its unit columns through the same stream
• Keep gLeft as a sparse row index built together with the U index instead of a dense GF256 matrix
• Add multi-row GF256 kernels that XOR or multiply-add many matrix rows into one destination in a single pass, addressed by bounds-checked row indexes
• Replace the column-wise U identity pass with a row-wise forward substitution that writes every row once, and check the unit lower-triangular U invariant while indexing
• Run back-substitution, gLeft*D and symbol generation through the multi-row kernels
• Stop materializing the D matrix: read symbol payloads directly as the first kernel source and drop the copy of upper D rows into C
• Rewrite Gaussian elimination in LU form: eliminate on A only, keep multipliers in place, and apply them to D with one forward and one back substitution per row
• Shrink the solver arena estimate now that D and the HDPC scratch matrices are gone
• Add NEON (arm64) and SSE2/SSSE3 (amd64) implementations of the multi-row kernels and the HDPC chain step, with a generic fallback for purego and other architectures and word-wise handling of tails shorter than 16 bytes
• Add golden digest tests pinning encoder output and decode results for 147 block shapes
• Add a regression test for the decode failure rate with reception overhead
• Add a randomized equivalence test of the LU solver against the previous Gauss-Jordan implementation
• Add multi-row kernel tests across lengths, source counts, misalignment and out-of-range row indexes
• Add benchmarks for encoder creation, repair symbol generation, 1400-byte symbols and 16MB blocks, and make Benchmark_EncodeDecodeFuzz fail on an undecoded block

v1.5.2

• Optimize allocs and GC

Commits

• 20e7198 Merge pull request #7 from xssnick/perf/solve-dataflow
• a3f7bc7 Row kernels: word-wise tails, no division in the bounds check
• e5a53f2 LU form for the dense system
• 83c04e7 Solve with row kernels: row-wise substitutions, no D matrix
• 6e8b743 Stream the HDPC products, keep gLeft as a row index
• 97709a0 Keep padding rows when extra symbols arrive, add golden digests
• 007ddfd Optimize allocs and GC
• See full diff in compare view

Dependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting @dependabot rebase.

---

Dependabot commands and options

You can trigger Dependabot actions by commenting on this PR:

• @dependabot rebase will rebase this PR
• @dependabot recreate will recreate this PR, overwriting any edits that have been made to it
• @dependabot show <dependency name> ignore conditions will show all of the ignore conditions of the specified dependency
• @dependabot ignore this major version will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself)
• @dependabot ignore this minor version will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself)
• @dependabot ignore this dependency will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself)

[dependabot]

Looks like github.com/xssnick/raptorq is up-to-date now, so this is no longer needed.