Note to people that design pages like these: If you're putting a blinking cursor on your page, I'm trying to type things. The first key I will press is the backspace key, if this takes me to the previous page, I now left your webpage and it's unlikely that I'll return. Might be just me though :)
"Only 1.1M" Why do I care? Is this framework designed for embedded systems or mobile apps? Smaller is better, but one MB vs 10 MB makes no difference for me as a guy writing server-side Java.
I suppose it makes a difference when it comes to getting your local dev environment set up. Plus, you might be able to get away with doing continuous integration that downloads all its dependencies every time it runs.
I'll trade a 1% reduction in implementation time for 10x the jar file sizes. I'll take "lighter" in terms of API surface area, memory/CPU usage, etc. though.
I was trying to figure out Java EE, and wrote some notes on how the various pieces fit together.
My final note was "You don't need this crap".
Oracle didn't take over Sun in 2009; the takeover started when Java EE started taking hold. It's given the cool little language Java a bad rap and I'm glad to see some efforts to bring Java back to its roots as a lightweight and easy language in the Web development sphere.
Java EE is not that bad. The problem is that it is for enterprise and you have tons of frameworks for interfacing with legacy mainframes and such.
Go for for EE web profile, which the lean version for pure web stuff.
All you really need is JPA, some ejbs and a front end(mvc framework or something) and that's it. Add in jax-rs, add some annotions to your ejb and you will have a rest api for you application for free basically.
IMO, JPA (and ORMs in general) is one of the bigger crimes-in-progress in Java. One of the things about Play I do like is that the Scala API eschews JPA.
EJB, however, manages to be even a bigger crime-in-progress than JPA. Which is a feat. Making immutable bean classes, and then providing a pleasing API for using them, is an exercise in futility. (And this may sound dogmatic, but I don't really care: immutable-by-default is the only acceptable practice. Embracing EJBs means you probably don't care about that, and that I want nothing to do with your code.)
Dude stuff goes in and out fashion faster than anything in tech. Orms went in, Orms went out, then in again(Django, rails) then out. Just need to stop being zealots following the latest trend and get things working.
People take the immutable route, then some other guy tells you your being silly. Trust me I've been programming long enough.
Besides you don't need Mutability for ejb. The majority of ejbs are just used to access the db.
No, thank you, I won't trust you on this. I've been a fairly loud critic of ORMs since the first time I was exposed to them (Hibernate, as it happens)--having seen them in practice hasn't changed that a bit. The conventional opinion on them may have changed, but mine never has; mutable data structures increase code complexity and greatly increase the surface area for state changes in places it wasn't (or shouldn't have been) intended.
And you may not need mutability for EJB, but it's almost always the case that you have it--it's partly a side effect of having to deal with people like you who consider it "not a big deal" and partly because JavaBeans are completely misused[1].
(For those following along at home, who might not be familiar with the culture of Java programmers, mutability in absolutely bizarre places eventually ruins your day at least once. See, for example, the Date[2] class, which for some very Java-y reason is completely and totally mutable. Questions related to the Date class are among my favorites for determining whether a Java programmer has any concept of mutability, and why "do you use Joda Time?" is a decent barometer of basic not-going-to-kill-your-project competence.)
.
Seriously, though: is "just get it working" the entirety of your defense for diverging from good practices? This isn't just me calling it "good practices," this is what Bloch says in head-shrinking detail in Effective Java and what pretty much everybody else who's had to write Java in anger eventually agrees with. This[3] isn't the best thing in the world on the topic, because I'm not going to quote Effective Java at length, but if you can read that and then come back with "just get it working," I'm gonna have to go light a candle for whoever inherits your code down the line because I've been there (still am there, in the day job) and it's hell.
I'm a functional programmer I know what immutable is. However if you've programmed a game in a functional programming language you know it also requires the code to become more complex in various situations. You have to introduce reactive programming to do it properly.
No longer can you simply change to coords of an entity to change a position.
In that situation I would recommend introducing mutability, trying avoid mutability would make it more complex.
Date api is one broken things about java. Everyone knows it is broken.
There about ten billion versions of good practise, I could quote them and say you are diverging from good practise?
I also recommend you keep things immutable, but not if it makes things difficult.
> if you've programmed a game in a functional programming language you know it also requires the code to become more complex in various situations
Irrelevant. A game is not a web application. A game has a claim to require mutability for performance concerns, but mutable state is generally safer in such a use case. A game is likely to have a few threads, each with a dedicated task or set of tasks, whereas a web application is not so constrained and any request may effectively duplicate another. Web applications that are cavalier about mutable state literally-literally invariably decompose to "synchronized blocks for all the things!" and, seriously, who wants to have anything to do with that?
> I can also find lots literature about following java bean convention as good practise. Your just cherry picking literature.
I'm not "cherry-picking" a Goddamned thing: I'm referring to one of the major, canonical sources of Java convention. You may find shitty Apress books that say "sure, use JavaBeans everywhere!", but that's why they're shitty Apress books. Effective Java is the book that literally-literally every competent Java programmer I have ever met has read and fully digested. There are areas of style where not everyone I've met agrees with Bloch, but immutable-by-default is very widely accepted except in areas (such as game development, though that is, again, irrelevant to the topic at hand) where performance becomes a legitimate concern.
But, except for those edge cases (and they really are edge cases, they certainly don't apply to web development) you are not going to find someone of significant credibility and stature in the Java community who is going to say "follow JavaBean convention". You just aren't.
Indeed, I too implement web apps
in a immutable manor when I can. Most of my ejbs are immutable, since they just add and return stuff. I don't like web servers relying on state. I also keep my ejbs mostly stateless using stateless ejbs funnily enough.
However when choosing JPA or JDBC - thus relying on boilderplate code loading data into java object constructors manually then I choose jpa.
It is a trade off for me. Boilerplate or keeping immutable data classes. For me that is a situation when adding immutability makes things more difficult.
And besides most well supported modern web frameworks like Django or rails also do it that way. So it is not really trade off when choosing EE / rails / Django. You have to have in most of those frameworks unless you rip out the orm.
I'm not really in a position to choose some obscure framework.
If you have a reason to use mutability in a specific case, then do it. If you default to mutability, then you are almost certainly writing poor code. And EJBs (bean classes in general) basically demand mutability to actually be usable.
I'm pretty sure "mutable-by-default is a major code smell" is not a very contentious statement in 2012.
Whenever I read about java things I get really confused. WTF is a super property, a BeanUtil, and a timeless JDateTime???
People like to hate on DHH and rails, but remember the "make a blog in 15 minutes" video. That's a compelling story for a web developer. Compare that to this.
If scrolling down gives me jodd experience, am I to assume that jodd is made for UX folks who care only about form and not function? That's what this webpage seems to be telling me.
That's a very good question (and one I'm personally interested in because I'm pretty frustrated with Play and the Play developers right now).
Jodd looks like a very 'old-school' Java framework. This is not a compliment--it looks like a few tools that sit on top of Servlets, and that's just terrible. Wire-up appears to be XML (bad developer, no cookie) and it just doesn't look better than any other Servlet-based library. (Personally, I am convinced that there are no good Servlet-based libraries, and this doesn't help change that.)
The various libraries they keep touting on their various pages are a little scary. I'm pretty well-versed in the Java ecosystem, but I don't know what "Petite", their dependency injection library, is. I know what Guice is, and I know what Spring (barf) is, but this thing is new, and by default shifty and untrustworthy.
"Props" - looks terrible compared to HOCON; gross syntax, no importing, etc.
Their database library touts itself as "better than JDBC", but looks almost identical and makes that little voice inside whisper, "it's a trick - get an axe."
I'm not trying to harsh on them, because they clearly spent a lot of time on this framework, but this is enterprisey as hell.
-Play 2 was released too early and Play 1 isn't getting any love anymore, so it's currently in an awkward spot. There's stuff you can do in Play 1 that can't be done in Play 2 yet; they're working on it, but it's not yet there.
-Play 2 is uncomfortable and vaguely user-hostile if you want to use Java. More than most Java apps, that is. It's fairly obvious that the Play team considers Scala to be The Way To Do It, so you'd better be on-board or you're probably going to hate using it. (I'm OK with Scala, though as I've said before I find it troublesome[1]. I hear eventually Kotlin is supposed to be supported though.)
-Play is all SBT. You may find that you love SBT, in which case this won't be a problem for you, but it has been a while since a build tool made me yearn for the days of ye olde makefile. There's a lot of "magic" involved and there's little coherent documentation for either SBT or Play's use of it.
-Play's DEV mode pins a core on my laptop and turns it into a campfire on my lap. A ticket asking what's up with this has gone unanswered.
-Play 2 is designed for single projects and trying to build sane multi-project systems (like, a frontend app, a common code library, and a backend app) can be an exercise in frustration. This has gotten less bad over time, but it's still not great. I literally don't remember the last time I wrote a web app that wasn't at least three separate pieces, so for me this is very nearly a game-breaker.
-The developers are apparently pretty insular. It seems like they are more than happy to have you fix their bugs, but adding features that you need--not so much. Sample size of one, so apply all necessary caveats, but I've found that offers of help and requests for guidance for making what you want to do fit into their plans for Play seem to get ignored[2]. I want to help make it better, so let's talk and see how to do that, yeah? (Lack of good communication isn't just a Play thing, but seems common throughout the Typesafe stack of libraries.)
-The community isn't really a community at all. If you look at their Google Group, it's mostly the same five questions over and over again and nobody seems to really stick around. I look at something like Rails--which I can't stand working with, but they have the "active, good community" part absolutely nailed--and you see a lot of newbie questions, but you also see a lot more people sticking to the community and, IMO, that bodes a lot better for the platform.
.
Honestly, Play isn't that bad, if your needs line up closely to the Play developers'. Mine don't, so for me it's a very frustrating experience. If you're just rapidly iterating on something small I have to think it'd be great - hot reloading works great (although compiling templates seems like it'd turn off designers) and I would believe that, once over the initial hurdles, you can probably Get Stuff Done very effectively. But every time I work in Play I start getting a little bit crabby and want to go write a better web framework (with blackjack! and hookers! in fact, forget the blackjack!) but I know I don't really have the time to do it justice. As such, my projects mostly end up being a set of libraries bolted together on top of Jetty (a handler, a regex dispatcher, etc.). Maybe at some point I'll codify them into an actual for-realsies "library". It'd be a good reason to learn how to do Java class reloading...
I was looking for actual information. Nice page and all, but it would be nice to see what Jodd brings to the table up front, rather than being distracted by lots of CSS/JS fx.
Take a look at the Play framework page. Starts with an elevator pitch and as you work your way down the page, you get more more details (without going into a lot of detail), enough to get a nice overview.
There's a time and place for heavy css/js effects, but I'm thinking this is not one of them.
