Use aggressive constprop for ^(::AbstractQuantity, ::Rational)
#487
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Codecov Report
@@ Coverage Diff @@
## master #487 +/- ##
==========================================
+ Coverage 83.77% 84.12% +0.34%
==========================================
Files 16 16
Lines 1344 1367 +23
==========================================
+ Hits 1126 1150 +24
+ Misses 218 217 -1
Continue to review full report at Codecov.
|
giordano
approved these changes
Oct 13, 2021
hustf
added a commit
to hustf/Unitfu.jl
that referenced
this pull request
Sep 1, 2022
* Use the `:fancy_exponent` IO context property to override the behavior of fancy exponents (PainterQubits#446) * Showing how to manually cancel units (PainterQubits#451) * Showing how to manually cancel units * Update conversion.md added the statement that `m/m` is automatically canceled * Remove contrived examples, fixed typo * More clear example This commit changes the expression from which units are shortened to 1km/2.5m, to make clear that the denomenator is part of the expression. It also fixed a typo (`NoUnit -> NoUnits`) * Shorten example It seems like some of the examples did not sit well with some of the contributers. I see their point completely, and in this PR I have made the example minimal and clearer. #LessIsMore * Added @ref, and more interesting example I added a reference to the NoUnits type (hopefully - I don't know how to chech that the reference actually works as expected). I also changed the example to the more complicated one, to make it less trivial and hopefully better showcase the usefullness. * Update docs/src/conversion.md Co-authored-by: Sebastian Stock <[email protected]> * As per the latest review, which got outdated Co-authored-by: Sebastian Stock <[email protected]> * fix example `isa(1m, Length)` (PainterQubits#454) `isa(1m, Length)` did not work after `using Unitful` but the corrected version works. * Elaborate on UnitfulRecipes in readme (PainterQubits#456) If I had known this would support plot axes with units, I would not have started down the road to PainterQubits#455. Maybe this elaboration will help someone else. [skip ci] * Add link to UnitfulBuckinghamPi in README (PainterQubits#442) I just registered the [UnitfulBuckinghamPi.jl](https://github.com/rmsrosa/UnitfulBuckinghamPi.jl) package. It solves for the adimensional groups in a list of Unitful parameters (quantities, units, dimensions, or combinations thereof). This PR adds a link to the package in the README. [skip ci] * Fix doctests (PainterQubits#464) * deg2rad and rad2deg with "units" (PainterQubits#459) * deg2rad and rad2deg * v1.9 * Update src/pkgdefaults.jl Co-authored-by: Mosè Giordano <[email protected]> Co-authored-by: Mosè Giordano <[email protected]> * Update README.md Mentioned NaturallyUnitful.jl * Update .gitignore (PainterQubits#482) * Update .gitignore * Update .gitignore Removed binary stuff [skip ci] * Use aggressive constprop for `^(::AbstractQuantity, ::Rational)` (PainterQubits#487) * Calculate correct `eltype` when multiplying `StepRangeLen` by `Units`. (PainterQubits#485) * Prevent promotion of StepRangeLen eltype when multiplying by unit * Release 1.9.1 * Fix multiplication of range and quantity (PainterQubits#489) * Remove unnecessary Bool-AbstractQuantity multiplication methods (PainterQubits#491) * made Bool-Quantity multiplication symmetric * added tests for Bool-Quantity multiplication * Remove unnecessary multiplication methods for Bool, AbstractQuantity Co-authored-by: Sebastian Stock <[email protected]> * Remove comment from previous Bool, AbstractQuantity methods Co-authored-by: Sebastian Stock <[email protected]> * Add NaN, -Inf tests for multiplication by false Co-authored-by: Sebastian Stock <[email protected]> Co-authored-by: Sebastian Stock <[email protected]> * Resolve method ambiguity (PainterQubits#495) * Fix fastmath trig functions (PainterQubits#497) * make `norm` use `norm` rather than `abs` (PainterQubits#500) * make `norm` use `norm` rather than `abs` * support broadcasting on ranges (PainterQubits#501) * Attempt to allow preferunits to work with non-pure units (PainterQubits#478) * Attempt to allow preferunits to work with non-pure units Should fix PainterQubits#457. Prior to this change, attempts to use something that looked like "preferunits(C/ms)" would result in strange behavior, without throwing any sort of errors. Now, that should work nicely. * Warn user when preferunits causes redundant units Issue a warning when preferunits creates redundant units, which could stop it from sucessfully simplifying certain quantities. * Fix new upreferred behavior for non-dimensional quantities * Add tests for preferunits changes * Update test/runtests.jl Co-authored-by: Sebastian Stock <[email protected]> * Fix preferunits tests * Apply suggestions from code review Co-authored-by: Mosè Giordano <[email protected]> * Fix issues added by removing excess arrays * No longer use string for key while checking for units of different scales * Update test/runtests.jl Co-authored-by: Sebastian Stock <[email protected]> * Apply suggestions from code review Co-authored-by: Sebastian Stock <[email protected]> * Update runtests.jl Co-authored-by: Sebastian Stock <[email protected]> Co-authored-by: Mosè Giordano <[email protected]> * Unit construction macro now defines docs for units (PainterQubits#476) * Unit construction macro now defines docs for units See PainterQubits#436 * Add documentation capabilities to most defined macros This lets all unit and dimension definition macros repect doc strings set in unit definition. Additionally, it automatically defines documentation for derived units using power-of-ten prefixes. * Add documentation for some units in pkgdefaults Documentation has been added for all base units and many other SI units, as well as adding a special case to the documentation so that kg and documented as being the SI base unit. * Allow documentation for log scales and units * Rewrite documentation format for all units through "Liter" Additionally, changed how the prefixed unit documentation is written; this now means that prefixed units made by other modules should be properly documented (tested with UnitfulAstro). * Added documentation for all remaining units Documentation is still required for log scales * Update src/pkgdefaults.jl Co-authored-by: Sebastian Stock <[email protected]> * Update src/pkgdefaults.jl Co-authored-by: Sebastian Stock <[email protected]> * Update src/pkgdefaults.jl Co-authored-by: Sebastian Stock <[email protected]> * Update src/pkgdefaults.jl Co-authored-by: Sebastian Stock <[email protected]> * Apply suggestions from code review Co-authored-by: Sebastian Stock <[email protected]> * Apply suggestions from code review Added descriptions of why certain alternate symbols are used (q for electron charge, minute for minutes) Co-authored-by: Sebastian Stock <[email protected]> * Remove excess newlines; add names to docstrings * Add NoDims documentation * Standardized number and equation format among quantity and unit docs * Standardize format of dimensions in unit docs * Apply suggestions from code review Various minor fixes and changes, mostly typos and format changes. Co-authored-by: Sebastian Stock <[email protected]> * Switch from inline to block code blocks in new docstrings. Also standardize docstrings to `2π` for 2 pi. * Add automatic documentation for everything defined by the dimension macro Also removed documentation from the pieces of log units that currently don't have any sort of automatic documentation generation system in place. * Fix documentation issues with two different symbols for vacuum permittivity constant. * Minor formatting fixes * Use block quotes for prefixed units' doc strings * Update src/types.jl Co-authored-by: Sebastian Stock <[email protected]> * Make automatic documentation optional Also (optionally) generate automatic documentation for derived dimensions. * Add dimension info in automatic prefixed unit docs. * Change method for adding dimensions to prefixed unit documentation Now no longer tries to `eval` stuff while running the macro. * Fixes to documentation and code structure for autodocs for various macros * Apply suggestions from code review Co-authored-by: Sebastian Stock <[email protected]> * Various format fixes and liters documentation changes. * Fix missing optional argument causing test failure * Remove no-longer-necessary NullLogger for liter and kilogram docs * Remove unnecessary autodocs arguments for non-SI-prefixed units * Apply suggestions from code review Co-authored-by: Sebastian Stock <[email protected]> * Fix invalid links being created in non-Unitful documentation * Add spaced out column for autogen parameter where possible * Apply suggestions from code review Co-authored-by: Sebastian Stock <[email protected]> * Update src/user.jl Co-authored-by: Sebastian Stock <[email protected]> * Add tests for autodocs * Add test for prefixed reference units * Hopefully fix docstrings and docstring tests on Julia-1.0 * Apply suggestions from code review Co-authored-by: Sebastian Stock <[email protected]> * More documentation changes from code review Also removed documentation ability for logunit and logscale (as these should be another pull request) * Apply suggestions from code review Co-authored-by: Sebastian Stock <[email protected]> Co-authored-by: Sebastian Stock <[email protected]> * fix range eltype (PainterQubits#503) * Fix some major invalidations to improve compile times (PainterQubits#509) * Fix some major invalidations to improve compile times Unitful.jl was flagged as a major invalidator of compilation downstream. Test: ```julia # From: https://timholy.github.io/SnoopCompile.jl/stable/snoopr/ using SnoopCompile invalidations = @Snoopr begin using DifferentialEquations function lorenz(du,u,p,t) du[1] = 10.0(u[2]-u[1]) du[2] = u[1]*(28.0-u[3]) - u[2] du[3] = u[1]*u[2] - (8/3)*u[3] end u0 = [1.0;0.0;0.0] tspan = (0.0,100.0) prob = ODEProblem(lorenz,u0,tspan) alg = Rodas5() tinf = solve(prob,alg) end; trees = SnoopCompile.invalidation_trees(invalidations); @show length(SnoopCompile.uinvalidated(invalidations)) # show total invalidations show(trees[end]) # show the most invalidated method ``` This method won the prize for the absolute most invalidations 🎉. But I think the bigger issue is that it simply doesn't follow Julia semantics. It fixes the types for issue PainterQubits#127 in a way that gives a stricter type than Julia would do in the normal cases (which is why the invalidation occurs). After this PR, heterogeneous arrays of numbers with Quantity in there act normally, and compile times are back to normal. Here's a showcase of it being normal: ```julia using Unitful, Test m = u"m" cm = u"cm" b = Union{Complex,Float64}[0 + 0im, 0.0] @test b + b == b @test b .+ b == b @test eltype(b+b) === Number b = Number[0 + 0im, 0.0] @test b + b == b @test b .+ b == b @test eltype(b+b) === Number b = [0.0, 0.0m] @test b + b == b @test b .+ b == b @test eltype(b+b) === Number ``` * Update promotion.jl * Release v1.10.0 * Fix PainterQubits#465 for isapprox with complex arrays (PainterQubits#468) * deg2rad and rad2deg * v1.9 * Update src/pkgdefaults.jl Co-authored-by: Mosè Giordano <[email protected]> * fix PainterQubits#465 * test, v1.9.1 * Update src/quantities.jl Co-authored-by: Sebastian Stock <[email protected]> * positive test * Update test/runtests.jl Co-authored-by: Sebastian Stock <[email protected]> Co-authored-by: Mosè Giordano <[email protected]> Co-authored-by: Sebastian Stock <[email protected]> * Fix `range` implementation on Julia master and resolve method ambiguities (PainterQubits#514) * _range(start, ::Nothing, ::Nothing, len) * _range(::Nothing, ::Nothing, stop, len) * Add tests for ranges with complex elements * Use modern syntax * Remove unnecessary `<:Real` * Fix range(start::Quantity{BigFloat}; step::Quantity{BigFloat}; length) * Fix for missing _rangestyle on 1.8 * Remove `real` * Use `step` everywhere (instead of `st`) * Fix range(;step, stop, len) on Julia master * Fix range(;stop, step, length) for mixed number/quantity * Add some newlines and comments * Rename _range_step_stop_length → _unitful_step_stop_length To avoid confusion with Base.range_step_stop_length * Refactor _range(start, ::Nothing, stop, length) * Rename 3-arg `Base._range` → `_unitful_start_stop_length` (`Base._range` always has 4 arguments) * Move promotion to `_unitful_start_stop_length` * Refactor _range(start, step, ::Nothing, length) * Resolve method ambiguity on Julia ≥ 1.7 * Simplify implementation * Use modern `where` syntax for `colon` * Add two-argument colon for dimensionless quantities * Remove unnecessary code related to encoding of μ (PainterQubits#511) * Add tests for encoding of μ * Remove unnecessary code * Use U+03BC everywhere * Same changes for U+025B/U+03B5 * Fix printing of `StepRangeLen` with complex elements (PainterQubits#513) * Enable `zero` for arrays with non-concrete eltype (PainterQubits#516) * Release 1.11.0 (PainterQubits#519) * Delete export of `convertr`, `convertrp` (PainterQubits#530) * Delete export of `convertr` There is no `convertr`, so don't export it. * Don't export convertrp It's not defined * Add `cispi` and `sincospi` (PainterQubits#533) * Add `modf` (PainterQubits#539) * Add `modf` for dimensionless quantities * Add tests for `modf` Co-authored-by: Sebastian Stock <[email protected]> Co-authored-by: Sebastian Stock <[email protected]> * Add link to UnitfulChainRules.jl to README (PainterQubits#542) * modified: Project.toml v1.10.1 reflect Unitful, add ConstructionBase modified: src/pkgdefaults.jl Use new macros, deg rad change modified: test/runtests.jl Add ConstructionBase * modified: Project.toml v1.11.0 Co-authored-by: Dilum Aluthge <[email protected]> Co-authored-by: KronosTheLate <[email protected]> Co-authored-by: Sebastian Stock <[email protected]> Co-authored-by: Jeff Fessler <[email protected]> Co-authored-by: Ricardo Rosa <[email protected]> Co-authored-by: Mosè Giordano <[email protected]> Co-authored-by: Mo8it <[email protected]> Co-authored-by: Samuel Buercklin <[email protected]> Co-authored-by: Oscar Smith <[email protected]> Co-authored-by: Alexander <[email protected]> Co-authored-by: Luke Bemish <[email protected]> Co-authored-by: Christopher Rackauckas <[email protected]> Co-authored-by: adkabo <[email protected]> Co-authored-by: Rashid Rafeek <[email protected]>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
^(::Dimension, ::Rational)and^(::Units, ::Rational)can be inferred on master. For^(::AbstractQuantity, ::Rational),@constprop :aggressiveis necessary. This fixes the test failures on master.