v2.2.0

SnoopCompile v2.2.0

Diff since v2.1.2

Closed issues:

• Use more than one branch for automatic PRs (#145)
• invaldiation_trees throws ERROR: unexpected loctag jl_method_table_disable at 3218 (#146)
• Using mocked functions in runtests.jl leading to ERROR: UndefVarError: not defined (#147)
• Too many coffee breaks while building the flamegraph (#150)
• Timing data: pairs vs tuples (#184)
• Remaining items for official "deep" release (#187)
• No method matching reprcontext (#193)

Merged pull requests:

• Add snoopl() to snoop time in LLVM optimization (#135) (@NHDaly)
• Expose changes to Julia via @snoopi_deep to report timing profile breakdown of all MethodInstances that are inferred during type inference (#139) (@NHDaly)
• Invalidations: support method deletion (#148) (@timholy)
• Fix isdefined check on nightly (#149) (@timholy)
• snoopi_deep: check that mi.def is a Method (#151) (@timholy)
• Add accumulate_by_method (#152) (@timholy)
• Fix tests on nightly (#153) (@timholy)
• Require method ownership (#154) (@timholy)
• Fix a rare test failure in snoopi_deep (#155) (@timholy)
• Don't leave the flamegraph headless in tests (#156) (@timholy)
• Dramatically improve flamegraph(timing) performance (#158) (@NHDaly)
• Add inference_triggers (#159) (@timholy)
• MassInstallAction: Install the TagBot workflow on this repository (#161) (@timholy)
• Add show for InclusiveTiming, add docstring (#163) (@timholy)
• Automated commit made by MassInstallAction.jl (#164) (@timholy)
• Qualify MethodInstances in tests (#165) (@timholy)
• Streamline ci.yml (#166) (@timholy)
• Simplify the module qualifier (#167) (@timholy)
• Add parcel for snoopi_deep (#168) (@timholy)
• Add tools for analysis of specialization (#169) (@timholy)
• Clean up temporary files in snoopl tests (#170) (@timholy)
• Option to consolidate consts in inference timing (#171) (@timholy)
• Add [compat] for Requires (#173) (@timholy)
• Improve parcel for deep (#174) (@timholy)
• Annotate flamegraph with isprecompilable (#175) (@timholy)
• Add edit support for inference triggers (#176) (@timholy)
• Add a few convenience methods (#177) (@timholy)
• Export InclusiveTiming and its constructor (#178) (@timholy)
• Allow flatten_times to return unsorted (#179) (@timholy)
• More ergonomics & cleanups (#181) (@timholy)
• More informative error for child-node itrigs (#182) (@timholy)
• Skip higher-order funcs for inference triggers (#183) (@timholy)
• Make tests more robust to timing order (#185) (@timholy)
• ascend integration and docs cleanup (#186) (@timholy)
• Switch from pairs to tuples in flatten_times (#188) (@timholy)
• Improve accuracy & speed of known_type (#189) (@timholy)
• snoopi_deep: return inclusive & exclusive times (#191) (@timholy)
• docs! (#192) (@timholy)
• Store full InferenceTimingNode in InferenceTrigger (#194) (@timholy)
• Support flamegraphs from children (#195) (@timholy)
• Fix a flaky test and add some precompiles (#196) (@timholy)