IMPORTANT: ClobberingReload.jl has been superseded by Revise.jl. Please consider using this package from 0.6 onward. ClobberingReload.jl is no longer actively developed (though pull requests are welcome).
ClobberingReload.jl helps with interactive development.
creload(::Module)is a drop-in replacement for
reload(modulename), that does not require rebuilding the state after
reload. The new code takes effect immediately, and works on existing objects.
@aimportare automatically reloaded when they are modified. This works as a successor to @malmaud's Autoreload.jl package.
no_warningsrun code with some warnings silenced.
See below for usage information, and the docstrings for details (eg.
Interactively (whether in the REPL, Atom, or IJulia):
using ClobberingReload using Houses # `using` modules is fine (unlike with `reload`) h = House(nwindows=10) println("Price of house:$(price(h))") > Price of house: 100 .... modify Houses.jl, change the `price` function .... creload(Houses) println("Price of house:$(price(h))") # no need to redefine h > Price of house: 130
NOTE: Parametric types cannot be defined inside a
creloaded module on Julia 0.5.
This is fixed on Julia 0.6. Parametric type aliases are still a problem, but there is an experimental
alternative that should solve this:
report any issues.
In IJulia (Jupyter notebooks),
creload will be called
automatically for modules that were imported using
whenever the module's source code has been changed.
using ClobberingReload using Images # regular using @ausing Foo # autoreloaded using @aimport Bar # autoreloaded import @ausing Car <: (Foo, Bar) # autoreloaded with dependency: whenever Car, Foo, or Bar # are modified, Car will be reloaded println(Bar.life_the_universe()) > 5 # ... modify Bar.jl, or one of its `include`d files println(Bar.life_the_universe()) > INFO: Reloading `Bar` > 42
The Julia REPL does not have execution hooks yet, but you can still trigger the autoreload feature for
@aimported modules by calling
areload() manually. Or you can use Revise.jl, which works around this issue by scheduling a background thread.
no_warnings silence some of
Julia's warnings. Typical usage:
scrub_redefinition_warnings() do include(filename) end
sinclude("foo.jl")uses the above code to run
include("foo.jl")without the usual redefinition warnings.
scrub_stderrcan scrub arbitrary warnings using regexes. See its docstring for details.
mod from scratch, creating a new module object,
then replaces the old module object with the new one. As a consequence:
import A st = A.SomeType(10) reload("A") st2 = A.SomeType(10) typeof(st) == typeof(st2) # false
st2 are actually of a different type, and cannot be equal. Functions
defined on the first
::A.SomeType will not work on the second, and vice
versa. This is inconvenient when working interactively.
ClobberingReload.creload solves this problem by never creating a second
module. It just evaluates the modified code inside the existing module object,
replacing the previous definitions.
using ClobberingReload import A st = A.SomeType(10) creload(A) st2 = A.SomeType(10) typeof(st) == typeof(st2) # true
reload cannot reload modules imported via
over 1 year ago