# Assertions and assumptions

**URL:** <https://discourse.julialang.org/t/assertions-and-assumptions/7389>\
**Category:** Internals & Design\
**Created:** [November 29, 2017, 4:57pm UTC](https://discourse.julialang.org/t/assertions-and-assumptions/7389 "2017-11-29T16:57:00Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![foobar\_lv2](https://avatars.discourse-cdn.com/v4/letter/f/ee59a6/32.png) [@foobar\_lv2](https://discourse.julialang.org/u/foobar_lv2)\
**Post date:** [November 29, 2017, 4:57pm UTC](https://discourse.julialang.org/t/assertions-and-assumptions/7389/1 "2017-11-29T16:57:00Z")

</div>

I find myself considering to use

```julia
@inline function __unreachable()
   Base.llvmcall("unreachable", Void, Tuple{}, Tuple());
end

@inline function assume(x::Bool)
    x || __unreachable();
end

```

This yields the nice

```julia
function foo(x)
    assume(x> 0)
    x>0 ? x: -x
end

@code_llvm foo(2.4)

define double @julia_foo_63669(double) #0 !dbg !5 {
top:
  ret double %0
}

```

Would this be a workaround for [https://github.com/JuliaLang/julia/issues/269](https://github.com/JuliaLang/julia/issues/269)?

That is, use the dreaded any for the data-layout and use assume instead of type-asserts in order to get static dispatch to the right method, without the branch?

Of course one would maybe like some macro-magic in order to turn assumes into asserts unless some command-line switch is provided; also I’m not sure whether one can force llvm to not actually emit code evaluating the conditional if it contains non-inlined function calls; in such cases, the assume should rather be eliminated.

A truly evil person might consider changing all asserts into assumes when running with --optimize, a la [https://github.com/JuliaLang/julia/issues/7732](https://github.com/JuliaLang/julia/issues/7732).

What do you think? Does an assume belong into base in order to tell the compiler about invariants you know?