Compile-Time DI vs. Run-Time DI

https://dimes.github.io/blog/compile-time-di-vs-run-time-di

by dimes • 7 years ago

47 36 7 years ago

dimes.github.io

_57jb• 7 years ago

Your point of breaking things as early in the process really is the winner.

Catching things early and often. Tooling that shifts things further left in the pipeline is my go-to default.

chris_wot• 7 years ago

Do you think it is rust that popularized this?

maxxxxx2• 7 years ago

Why Rust? Strongly typed languages had this capability for a long time.

chriswarbo• 7 years ago

It's been standard programming dogma for decades, e.g. https://developers.slashdot.org/story/03/10/21/0141215/softw...

kangnkodos• 7 years ago

DI is Dependency Injection.

stabbles2• 7 years ago

What this blog post does not address is the joy of configuring DI with XML files /s

Isn't it bizarre how software developers have taken runtime DI to such extremes that it was considered a good idea to configure services in XML or YAML -- basically another language, that has to be parsed, can be malformed, etc?

duality• 7 years ago

The creators of Guice mention how configuring the dependency graph in Java was a big advantage it offers over older XML-based frameworks like Spring.

douglasisshiny• 7 years ago

Spring has allowed for java configuration for some time.

vips7L• 7 years ago

I've never configured Spring DI via XML and I've been a Spring dev for at least 2 years. It's always been via Java annotations and config classes.

lmm• 7 years ago

Early Java took a stupid decision to deliberately restrict expressiveness in the name of being "blue collar" (ironically enough Go largely retreads the same mistake). The result was a language so cumbersome that people would do anything to avoid having to write actual Java. Stupid policies at large companies that subject "configuration" changes to much less scrutiny than "code" changes compounded the problem.

The decision to use XML was actually quite reasonable for many developers given the constraints at the time. Thankfully Java has mostly seen the light these days, adding generics, first-class functions, and good-enough multiple inheritance (though it still has very restricted and cumbersome metaprogramming - classes are not first-class, and the lack of HKT makes it very difficult to work with functions generically since you can't abstract over arity), and people are gradually realizing that this different environment warrants different choices. (Companies still have terrible policies about code versus "config" though).

dullgiulio• 7 years ago

I never understood this. DI is for deciding program behavior. That should be a user choice. I don't understand how final users are supposed to (1) write XML files instead of providing simple human readable CLI flags, (2) know the program internals and being able to type in fully qualified class names.

$ ./my-server -reporting=email

setting com.mycompany.package.subpackage.di.providers.abstract.ReportingProvider to com.mycompany.package.subpackage.di.providers.impl.EmailReportingProviderImpl.

The whole DI is a wrong concept. Dependencies should be instantiated at top level (main function) and explicitly passed down the call graph as interface instances.

DI breaks the call graph with (configurable) side effects. If you pass down dependencies explicitly you don't even need mocking frameworks for testing.

spand• 7 years ago

You are just refuting your own straw man.

Passing dependencies also qualifies as DI and is for the benefit of the developer. Maybe you are only talking about DI frameworks or containers in which case I share some skepticism but in general the concept of DI is pretty sound.

jcelerier• 7 years ago

Isn't it bizarre how software developers have taken runtime DI to such extremes that it was considered a good idea to configure services in XML or YAML -- basically another language, that has to be parsed, can be malformed, etc?

well, then how do you do it ? what do you do if your client / boss tells you "I want to configure the plug-ins & classes availables in the software through some file editing without having to recompile anything" and your codebase is in C++ / Java / Go / D / Rust ...

JetSpiegel• 7 years ago

what do you do if your client / boss tells you "I want to configure the plug-ins & classes availables in the software through some file editing without having to recompile anything" and your codebase is in C++ / Java / Go / D / Rust ...

You compile all the (tested) combinations as static executables and shove them all into the release package.

stabbles2• 7 years ago

Yes, in that case you probably need to parse a config file.

But you might also encounter DI config files for specific environments, like development where you inject MockMailer, TrivialLogger and InMemoryDatabase instead of RealLifeMailer, CloudLogger and ActualDatabase.

kazagistar• 7 years ago

Java has a fairly nice compile time DI framework in Dagger. However, it feels like it is missing a compile time version of the rest of the web stack: compile time routing and compile time template compilation, for example. Does anyone have any suggestions to fill out this stack?

snuxoll• 7 years ago

It’s not perfect by any means, but there is Ktor as an example. Use kotlinx.html for your templates and everything is defined at compile time.

benbjohnson• 7 years ago

I don't understand why you need a framework for dependency injection in Go. I've written Go for years and I just use the `main` package to instantiate and wire up everything. It's really straightforward, easy to debug, and there's nothing to generate. It's all type checked at compile time.

I'm not trying to rain on the author's parade. I genuinely don't understand the benefit—especially in relation to the complexity of adding an extra dependency and layer to my application.

strulovich• 7 years ago

How many active contributors and lines of code does your projects have?

I’ve found that the advantages for DI frameworks (besides encouraging more unit tests and enabling them) to be more valuable for bigger and messier projects.

sagichmal5• 7 years ago

I find precisely the opposite, that as a project gains (and loses, and gains again) maintainers, the indirection and mental overhead that DI frameworks necessarily bring to the table vastly outweigh whatever perceived benefit there was in introducing them.

Conversely, a declarative, step-wise func main, with components manually constructed and passed to each other as dependencies where necessary, is sometimes tedious, but never confusing, and always refreshingly easy to maintain, no matter how much time you've spent with the project.

dvlsg2• 7 years ago

I've found containers are helpful when you want to replace one or two specific pieces in the middle of the dependency tree for testing - let's say a mock database, for example.

On the other hand, I've also found myself avoiding writing code which requires deep mocks like that in order to test. Functional core, imperative shell helps with that.

eeperson2• 7 years ago

The benefit of a dependency injection framework is that you don't have to figure out the order of the dependency tree. If you do this work manually, then you have to make sure to instantiate things in a particular order so that dependencies are initialized before the classes that depend on them. This can be big pain if you have a large number of classes and start refactoring things.

I can understand some hesitation about using a dependency injection framework. Some of them can be very hard to reason about (especially the run time frameworks). However, they don't have to be that hard to reason about. My personal favorite is a compile time dependency injection framework for scala called Macwire[1]. In scala, you might do manual dependency injection like this:

  val instantiated = new Class(dependency1, dependency2)

With Macwire you could write the same thing as:

  val instantiated = wire[Class]

The wire macro above, automatically looks up dependencies by type and creates an instance of Class with those dependencies as arguments. If you combine that macro with Scala's built in support for laziness then you don't have to worry about initialization order either. This gives you something that is fairly easy to reason about (it is basically what your were doing by hand before) and will sort our necessary dependencies for you (and error out of they are missing).

[1] https://github.com/adamw/macwire

sagichmal5• 7 years ago

The benefit of a dependency injection framework is that you don't have to figure out the order of the dependency tree.

What! This is essential knowledge. Without it, how can you possibly build a coherent mental model of a program?! I’ve never heard this justification before, it’s bonkers.

Compile-Time DI vs. Run-Time DI

Related Stories