Whereas in the modern JS world, you start out with a lean project and then add small libraries from NPM that all work slightly different for every feature you need, in the Spring world where the framework has matured for 10+ years most of the things you will need are in the box or are available as external libraries that all work on the same defined interfaces.
Spring Boot with Java 11 or Kotlin are still my preferred backend stack for that reason. It’s solid, has aged well and makes me super productive.
Until it doesn't, then everything goes to Hell very quickly.
Spring is close to impossible to debug, because everything is driven by annotations, configuration settings, property files, beans defined in XML, default values, or code that runs because you added a dependency even though you never explicitly call it.
Program logic is defined in anything and everything, except actual Java code.
The whole thing is a nightmare ignoring every principle of simplicity, composability, referential transparency, and every other established principle of good software design.
I see a comment like this every week on HN and I've been meaning to write a blogpost or two to address this.
Yes, as you mentioned there is a myriad of ways to configure a Spring(Boot) application, and I think that is one of things that make it such a powerful framework or platform to build a service on. To address a few of your points:
* beans defined in XML : we don't do that anymore since annotations got introduced. I haven't written a beans.xml file in years.
* Program logic is exactly where you expect it: in the Java code. All the property files and annotations are used for configuration to allow for different environments. This setup allows us to run the exact same binary jar or docker image built from CI to run in CI, in dev, in staging and in multiple different production environments.
* configuration settings and property files are the same thing as the configuration values go into property files, you can have multiple property files to group configuration logically by their purpose, and if needed duplicate each property file per environment to group those together. I know my database config for the staging environment is in database-staging.properties. If that is not enough you can easily override any of the values with environment variables
* Debugging : another point where Java gets huge flak here on HN is huge stack traces, well I think those are actually helpful in figuring out exactly what is running and giving you trouble in a specific environment. And because of the separation of configuration from code mentioned above it is trivial to setup a testcase with the exact same configuration that you are having trouble with and use a debugger on that testcase to inspect and debug.
* Simplicity and composability : While the framework can look complex if you don't understand the configuration system, it is actually there to facilitate composability. See my point about running the same binary in all environments. The big idea is that you put all code in the same repo and then compose and configure your application at runtime with the configuration system.
> beans defined in XML : we don't do that anymore But you still can do that! Also you still can do DI with Autowired private fields instead of injecting via constructors. You can also create beans based on properties. You can also create beans automatically via a included JAR on your class path. On Spring Boot you can also have beans created based based on some conditional annotation. And now there's yet another new DSL [1] to create your beans programatically on startup time. Tracking down and debugging what's loaded in Spring is super hard compared to Guice for example.
Configuration is another piece of software that I have a really hard time to think why they do that. You can configure via a YAML file, AND via a properties file, AND overriding via environment variables, AND also with a configuration server that store your properties on a separated server, and they are all mixed together in a weird single configuration entry for the specific environment that you have. Probably this made sense some 10 years ago when we loved J2EE but why support this kind of madness nowadays?
What I feel Spring is missing so much is forgetting about all the past and where the framework came from, start from scratch with simple premises around reactor-netty [2], only supporting the functional DSL, no crazy class path scanning, no crazy configuration loading, just a simple DI framework with some batteries included.
[1] https://spring.io/blog/2017/08/01/spring-framework-5-kotlin-... [2] https://github.com/reactor/reactor-netty
Even stuff in “core” dependencies like webpack/yarn/npm/npx/nvm/babel changes a few times a year.
Right now doing a react native app and a browser extension. While javascript/typescript is fun to write, the time you have to spend fixing the tools is just horrible. Whiping cache, rebooting, switching package versions.. It makes the overall experience cumbersome.
I would not call something stable if it manages to work 2 whole years.
There was definitely an era of that, but over the past decade or so it's been acknowledged as a problem, and there's been a lot of effort put into making things simpler and more vanilla. Modern Spring is much much closer to plain old code. (Of course it's still possible to write in the XML/setter injection/afterPropertiesSet style, and always will be - that's backward compatibility for you - but it's very possible to migrate existing applications to a better style in place, which very few platforms manage).
> Same with maven/gradle, which turns what should be a developer-focussed build system into an enterprise XML mess.
Maven is pretty close to the perfect build system, it's just taken a while for knowledge and practices to catch up with it. All the stuff people love and praise Cargo for, Maven has been doing for 15+ years.
However, I don't want to sound too negative. I can see a perspective for the JVM (Graal/Truffle) as a polyglot runtime where most of the ancient Java stuff is hidden from you, and you're using JavaScript or other non-JVM-only language as a migration strategy to get rid of Java alltogether in the distant future.
Could you be a bit more concrete? I don't recognize any of your descriptions in my experience of maven.
There are probably some bad plugins out there, but as in any system the solution is not to use them.
The complete lifecycle is listed in the documentation, and it's all very common-sense.
pom.xml isn't meant to be Turing-complete, and if you try to use it like a Turing-complete build language you will get yourself in trouble (as many of the early Ant diehards did). You need to treat it as a declarative config file, no different from e.g. Cargo.toml.
> I just don't have the patience with wading through crap that 2003ish Java nerds thought would be a good idea. And increasingly, as Java libs get EOLd, this will be a major problem for monolithic Java apps with tens of submodules. I don't think you can expect younger devs to keep maintaining daddy-o's web framework crap and cleanup the mess left by Java heads.
It's just the opposite. I don't expect anyone to want to maintain Ant or Gradle builds in the future, since they can have any kind of crap in them. But Maven builds are forced to be kept sensible; the tool just isn't flexible enough to let you put crap in your build definition, by design. (If you really insist on doing crap you can make a crappy plugin, but your plugin is then forced to be versioned, tagged in VCS etc., so stands a better chance of being picked up by your organisation's normal code quality efforts).
There's a jetty-maven-plugin? What for? When I use jetty I treat it as a library, embed it in my application, write my own main() and then my application is just a normal application.
> servlet api versioning conflicts
> multiple mvn central listings for the same artifact
These are a genuine problems but they're ones that every mature ecosystem struggles with. Maven has better tooling than most IME - you can use the enforcer plugin to require dependency convergence, and when multiple packages contain the same API you can put exclusions in a common parent. No ecosystem has really solved the problem of package renames nicely, IME.
> ad-hoc scripting using ant plugin
Yeah don't do that. If people will insist on shooting themselves in the foot, there's only so much you can do to stop them.
> shading/fatjar plugin
What went wrong there? IME the shade plugin is very good. There are a couple of inherent JVM-level limitations (having to specify transformers for service locator files etc. because of the way the JVM spec is written) but it's hard to imagine any other build tool could do better.
> fragile local dependency resolution
Not my experience at all - what specifically goes wrong?
> dozens of pointless submodules
Submodules are cheap. IME when coming to work on a new project it's much easier to understand the build/dependency structure if there are lot of submodules rather than a complex build - e.g. rather than trying to have a custom build phase for generating some RPC stubs or whatever, it's fine to just have them in their own submodule and then it's obvious to the reader what the dependency graph looks like. What's the downside?
> A customer of mine even had a solid 3 week outage when they wanted to setup a product for blue/green deployment based on maven.
What? How?
> Makes autotools look like a sane build tool.
Strongly disagree with that one. Autotools has several layers of template/macro expansion that all seem to do much the same thing (so why not just one?) and encourages builds that use arbitrary unversioned command line tools from the host system, meaning you very rarely get reproducible behaviour on a host with a different setup. Maven gives consistent, declarative builds - what's not to like?
Modern Spring, as far as I can tell, is still very much driven outside of Java code, it's just migrated from XML to annotations and yaml properties files. Seems almost like the typical Spring application has more lines of annotations and properties files than lines of Java code.
I am fond of Maven, though, especially with Gradle (superior to pom files).
Disagree; there's been a shift to using constructor injection over setter injection, which often obviates the need for Spring-specific lifecycle methods. So it's now much more common to have a service class that, while it may have Spring annotations on its constructor, can also be used in a first-class way in plain old Java, and e.g. doesn't use Spring for unit tests. If you're using Spring at all then part of your application wiring lives outside Java code almost by definition, but it's now more common to express wiring logic as e.g. an annotated Java method rather than encoded into XML. It's not a complete departure, and it may not go as far as you or I would like, but I do think there's been a real cultural shift.
> especially with Gradle (superior to pom files).
It really isn't. Arbitrary unmanaged code in the build definition is great for writing but terrible for maintainability.
Despite Spring being Enterprise it still scales down to simpler apps pretty well due to its modularity.
These can be build into a single jar app that is easy to build and deploy, and combined with a database migration library, maintain their own database schemas. Very, very usable.
Yes, I probably would not use Java on the frontend unless it was for a technical audience.
People complain about the verbosity of Java, but how much time do you spend learning a new framework and all of the gotchas? At the end of the day it seems easier to just write the boilerplate. All of that verbose code means you got exactly what you asked for.
JVM projects never surprise me. When there is a problem debugging is a dream.
That. I've used the Servlets API for years in the early to late 00s.
I'm now (for the past 5 years) using Express with Node.
Static typing and cosmetic differences aside, Servlets and Express (or Node's raw handlers) are pretty much the same API.
Request, Response objects, handlers, middleware (iirc we used to call those last "filters" in the Servlet era), etc. It's not like it's some alien convoluted design.
(Java Server Faces now was another thing)
But how do you even know what code to read when so much of the program's logic is in annotations and other kinds of action at a distance?
It's not that Spring is badly coded or something, I just don't need these surprises and the magical thinking surrounding newbie Spring Boot projects.
Nothing.
All the carefully crafted validation annotations just don't do anything, no validation is performed, and there is no indication whatsoever of this.
I had a problem with security that I could only solve by adding an "Order" annotation to my class. Reading the code and stepping through it in the debugger did me absolutely no good.
Could you elaborate on the overengineering part? There are many modern frameworks that are much lighterweight than Spring/Spring Boot, and with the introduction of lambdas in Java, things are even more straightforward.
Also, gradle doesn't use XML.
All our customers doing Java Web projects are either on JEE, or a CMS platform running on top of servlets and JEE related JSRs, like Liferay and AM.
In 2006 I played around with Jackrabbit for an architecture concept (for managing airworthiness of airplanes and kit). That was before the Iphone even existed. I was honestly surprised Adobe is still using it in their AEM product. From what I recall, it manages a document tree, based on a 2004-ish, Drupal-like node graph in typical Java enterprise fashion.
I'm no insider, but my guess is Adobe is struggling for some time now to integrate AEM, Magento, and other purchases into a lineup for creating shopping experiences. We'll see how this works out.
Not any more than PEPs are.
I have learned it pays off longterm to watch caravans pass by.
Pure HTML and CSS with minimal JavaScript rules rendering performance.
The flip side of Java's stability is stagnation. On the server-side, customers use mostly Spring/Spring Boot these days to make Java's overengineered stack usable, using even more over-engineering and metaprogramming. Same with maven/gradle, which turns what should be a developer-focussed build system into an enterprise XML mess.
SWT? I'm glad it works for you. Last I heard, IBM had pulled out of developing Eclipse and SWT/RCP over ten years ago. In the one project where we used it, the app looked like an IDE, and customers hated, then cancelled it. Designed as wrappers for OS-native GUI controls, it has even less of a place in a browser than Swing had. Last I looked, of the Java rich-client frameworks, only Swing has been updated with HIDPI support, whereas JavaFX and SWT haven't. None is using any of those (with the exception of experimental JFX apps) for new apps in this decade.