v0.2.0

DataFlowTasks v0.2.0

Merged pull requests:

• Add debug mode to handle errors in tasks (#3) (@ffevotte)
• Add tests for errors handling in tasks (#4) (@ffevotte)
• extract TiledFactorization module (#8) (@maltezfaria)
• New surface API (#9) (@ffevotte)
• New version of logger (#10) (@maltezfaria)
• Improve performance of dataflow analysis (#13) (@maltezfaria)
• Test TiledFactorization in CI (#17) (@ffevotte)
• Makie Scene (#18) (@Vincent-Sivadon)
• add a ReentrantLock to the DAG structure (#19) (@maltezfaria)
• Allow access mode tags in macro parameters + more macro tests (#20) (@ffevotte)
• Documentation (#22) (@ffevotte)
• Optional dependencies management (#26) (@ffevotte)
• Don't change logging/debug/sequential mode unnecessarily (#28) (@ffevotte)
• support for v1.6. Fixes #29 (#31) (@maltezfaria)
• add method to restart scheduler (fixes #24) (#32) (@maltezfaria)
• Changes prior to first release (#34) (@maltezfaria)
• Add a new blur-roberts example (#36) (@ffevotte)
• Improve the critical path implementation (#37) (@ffevotte)
• Minor improvements (#38) (@maltezfaria)
• Transition from Requires.jl to v1.9-style weak dependencies (#42) (@ffevotte)
• Update compat bounds (#44) (@ffevotte)
• Improve tests with Aqua.jl, simplify AbstractArray interface, and update CI (#45) (@maltezfaria)
• New implementation of weak dependencies handling (#46) (@ffevotte)
• Autoformatting (#50) (@maltezfaria)
• Changes to API (#52) (@maltezfaria)
• Generate examples notebook in the CI process (#55) (@maltezfaria)
• Fix second example in the doc (#56) (@ffevotte)
• Simplify README.md generation (#57) (@ffevotte)
• New example: Longest Common Subsequence (#58) (@ffevotte)
• Logging should wait for task graph to be empty (#61) (@maltezfaria)
• Expose ExtendedLogInfo and improve logging experience without add-ons (#62) (@maltezfaria)
• Simpler describe(loginfo, categories) API (#64) (@ffevotte)
• Update examples to use the describe(log_info) API (#65) (@ffevotte)
• CompatHelper: add new compat entry for TOML at version 1, (keep existing compat) (#66) (@github-actions[bot])
• New example: merge sort (#67) (@ffevotte)
• Rename DataFlowTasks.@Spawn to @dspawn, export it (#69) (@ffevotte)
• Update Cholesky example (#71) (@maltezfaria)
• Bump actions/cache from 1 to 3 (#72) (@dependabot[bot])
• Bump peter-evans/create-pull-request from 3 to 5 (#73) (@dependabot[bot])
• Bump actions/checkout from 1 to 4 (#74) (@dependabot[bot])
• Expand discussion on nesting tasks and limitations (#77) (@maltezfaria)
• Add compat entry for stdlibs (#78) (@maltezfaria)
• Allow for Tasks as a data-dependency in a DataFlowTask (#80) (@maltezfaria)
• Implicitly add a "data dependency" between each DataFlowTask and its underlying Task (#81) (@ffevotte)
• Documentation updates (#84) (@maltezfaria)
• Use upstream version of reverse iterator for OrderedDict (#85) (@maltezfaria)
• Bump version to 0.2 (#86) (@maltezfaria)

Closed issues:

• New logger (#1)
• Better handling of errors (#2)
• Make TiledFactorization into its own package + clean-up dependencies (#5)
• Improve the API (#6)
• Better tests (#11)
• Improve the data analysis algorithm/implementation (#12)
• Simplify and improve critical path algorithm (#14)
• Plot recipes (#15)
• IO during task insertion causes race condition on DAG? (#16)
• Decide where to put PseudoTiledMatrix (#21)
• Dependencies not taken into account (#23)
• DAG environnement not beeing cleaned (#24)
• Tricky behavior with captured variables (#25)
• Useless code invalidations with enable_log() & co (#27)
• Add support with v1.6 (#29)
• Explain figures on README which are not generated by the CI (#30)
• Tests on CI take too long (#33)
• Logging interface (#35)
• Rendering information about LogInfo (#39)
• Document troubleshooting and debugging strategies (#40)
• Towards v0.2.0 (#43)
• Make sure @using_opt works as expected (#47)
• Ensure proper formatting with JuliaFormatter.jl (#48)
• API change / simplify internals of DataFlowTasks? (#49)
• Run again the Cholesky performance comparison (against MKL) and make sure it is still relevant. (#53)
• Example does not work (#54)
• Thread-safety of logging? (#59)
• Logging should wait for task graph to be empty (#60)
• Revert to DataFlowTasks.@dspawn (#68)
• Update Cholesky example with LoopVectorization kernels (#70)
• Fetching inside a DataFlowTask (#79)