ryan/libdpf
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Checkpoint the party/runtime stack before share-program and malicious-mode work.

Browse source 
Ship the TLS mesh, composer, Beaver/Yao/leaf MPC, prep/online paths, apps, and docs so the tree is pushable before elevating share_expr, security_mode, and prep resume.

Co-authored-by: Cursor <cursoragent@cursor.com>
This commit is contained in:
Ryan Henry 2026-09-28 05:59:19 -06:00
parent 695f8e84f7
commit 0d22946a0e
1835 changed files with 170291 additions and 2849 deletions
Download patch file Download diff file Expand all files Collapse all files

4
include/dpf/sequence_memoizer.hpp
View file

@ -289,6 +289,7 @@ struct pointer_facade
/// alternate levels.
/// @tparam DpfKey DPF key type
/// @tparam Allocator allocator type
/// \complexity One buffer of `recipe.num_leaf_nodes()` interior nodes, reversed in place each level.
template <typename DpfKey,
typename Allocator = aligned_allocator<typename DpfKey::interior_node>>
struct inplace_reversing_sequence_memoizer final
@ -387,6 +388,7 @@ HEDLEY_PRAGMA(GCC diagnostic pop)
/// intact. Default workspace for `eval_sequence` on a recipe.
/// @tparam DpfKey DPF key type
/// @tparam Allocator allocator type
/// \complexity Two buffers: `2 * recipe.num_leaf_nodes()` interior nodes.
template <typename DpfKey,
typename Allocator = aligned_allocator<typename DpfKey::interior_node>>
struct double_space_sequence_memoizer final
@ -442,6 +444,7 @@ HEDLEY_PRAGMA(GCC diagnostic pop)
/// @brief Every level of the recipe's traversal.
/// @tparam DpfKey DPF key type
/// @tparam Allocator allocator type
/// \complexity `level_endpoints().back() + num_leaf_nodes()` interior nodes, the recipe's stored levels plus the leaves.
template <typename DpfKey,
typename Allocator = aligned_allocator<typename DpfKey::interior_node>>
struct full_tree_sequence_memoizer final
@ -506,6 +509,7 @@ auto make_sequence_memoizer(const sequence_recipe & recipe)
} // namespace detail
/// \complexity O(n k) interior traversals in the worst case and O(k) node workspace. k is the number of listed points and n is `depth`. The breadth-first buffer is 2k nodes, so each level traverses at most one node per point. Shared prefixes do fewer traversals. A recipe memoizer instead stores O(recipe leaf nodes) (see that memoizer).
HEDLEY_PRAGMA(GCC diagnostic push)
HEDLEY_PRAGMA(GCC diagnostic ignored "-Wignored-attributes")
/// @brief One-level sequence workspace bound to `recipe`.