Build History for branch/main

Regroup MS_queue and spsc_queue under their own subdirectory. (#151)

Fix peek -> peek_exn in doc. (#148)

Prepare for release 0.5.0 (#147)

Michael-Scott queue : safe - unsafe versions (#146) Optimized safe and (more optimized) unsafe MS queue with tests running for both.

Using multicore-magic-dscheck for dscheck tests (#143) Add multicore-magic-dscheck to write dscheck tests on a data structure using Multicore-magic.

Merge pull request #142 from lyrm/skiplist-bench Add back an accidentally dropped line.

Tests on 5.2.0 on Windows (#141)

Optimize spsc queue : unsafe optimizations (#135) Applied optimizations: - Add padding with Multicore_magic.copy_as_padded to avoid false sharing - Use `Atomic = Transparent_atomic`, `fenceless_get`, and `incr` - Avoid indirection in array - Avoid float array pessimization - Add a simple `float` test and prevent inlining

Merge pull request #140 from lyrm/skiplist_test_dune Fix cycle dependency in skiplist tests

Require `sherlodoc` for search bar This requires `dune` 3.14, `odoc` 2.4.1, and `sherlodoc` 0.2 so that generated documentation should include the new search bar.

Link to benchmark results on `pascal` The results on the `pascal` machine are likely to be more representative of relative performance than on the older Opteron based `fermat` machine.

Fix some typos, English idioms, and reformat the doc comments As a stylistic thing, I would recommend using an extra space after a period. Like this. And this. This makes it easier to see where sentences begin and end when using monospaced fonts and makes the text easier to read. The convention of an extra space after a period is also recognized by the `fill-paragraph` command of Emacs.

Merge pull request #133 from lyrm/optimize-spsc-queue-safe Optimize spsc queue : optimizations without `Obj.magic`

Bench on 5.2.x

Prepare for `summer`

Add a benchmark using work-stealing deque as a SPMC queue This is a technically possible use case of a work-stealing deque and also a possible, although non-ideal, pattern for some programs running on a work-stealing scheduler.

Upgrade to multicore-bench 0.1.2 The Stdlib benchmarks now live in multicore-bench.

Fix MS queue test loop loop without safe point

Upgrade dscheck to allow testing on OCaml 4 with ocaml-ci

Avoid code duplication in Treiber's stack using a GADT

Minor optimization to `Size` There is no need to mask the counters outside of snapshotting.

Rewrite or refactor benchmarks using multicore-bench

Lock free skiplist with size (#99) Add skiplist with generalized size computation --------- Co-authored-by: Carine Morel <carine@tarides.com>, Sooraj Srinivasan <cs20b075@smail.iitm.ac.in>