next »

[mcc]

Yeah, I was talking about JavaScript+HTML5 vs C++ or Flash vs C++ on different devices. "Language to language to platform" compilation adds definetly a big layer of complexity. This is why it has to be well made to hide the complexity... At least GWT/Java to javascript works well enough that big companies are using it.
GWT development is IMO really comfortable. Integrating special javascript stuff is pretty easy. I once worked on a chrome addon with GWT, it was really a nice experience.

One thing that would definitely worry me is, ok, maybe the compile step works, but how likely are you to be able to run in a debugger in any way that makes sense?

[Chromanoid]

Debugging works pretty well. You can more or less debug your application like a normal Java application (server and client at the same time) step by step in Eclipse... This is because you can run the application in "hosted mode". This means the thing doesn't run in javascript but on your jvm that only sends every action to your browser. I wonder if this works for Flash and PlayN too. See http://code.google.com/webtoolkit/doc/latest/tutorial/debug.html and http://code.google.com/webtoolkit/doc/latest/DevGuideCompilingAndDebugging.html#DevGuideDevMode

[Serapth]

One thing about PlayN from my initial exposure is that the documentation is almost non-existent and the installation process is near to brutal, especially if Java is not your primary language.

If you aren't extremely confident in Java/Eclipse/Maven/Git, I put together this installation guide which is extremely detailed and should get you up and running.  I am going to take a deeper look into PlayN and possibly do a series of tutorials, as like I said, the documentation is extremely lacking.

I don't know if PlayN will take off or not, it's an interesting idea, but it does rather feel like Google doesn't know its target audience.  For a game library getting it up and running really gave me flashbacks to enterprise development and working with companies like IBM and EDS.  Given their target market is Indie and small developers, this just seems really silly.

[Mikademus]

Excellent guide! And an astonishing read! I was thinking "how could they?!" and my eyes kept widening as I read! Yeah, that nebulous complexity has to be done away with or there isn't a chance for PlayN to take off. Coupled with poor documentation and an unclear purpose or actual advantage things seems to talk against it atm.

[Chromanoid]

Thank you for this nice guide You are right it seems very complex, Java developers usually have done most of your tutorial already and are used to maven, but if PlayN really wants to target javascript, iOS and Flash developers, stuff like this and dozens of plugin installs are a no-go.
I think the main target groups of PlayN are Java, Android & GWT developers as well as medium-sized game companies and social game developers.

[Serapth]

Yeah, truth of the matter is, someone exposed to Java development ( and Eclipse in particular ) will look at that guide and go, meh, that's not so bad.  For someone experienced with Java, Maven, Git and Eclipse it really is pretty typical, the process of working on Android is remarkably similar.

The problem is, Google are targeting this at HTML5 and Flash developers and in that case, learning all those other technologies just adds to the learning curve.  They are really going to have to streamline the process, like both of you said, if they want to appeal to non-Java developers.

At the core though, I think a lot of the complexity is attributed to Eclipse.  I know there are a core of people that absolutely love it, but I generally find it a pain in the ass.  When I was working on Android last year the environment corrupted itself so many times I lost count.  Then even when working on this tutorial, I started from a scratch install and Eclipse latched itself to the wrong java version ( the JRE instead of the JDK ).  This isn't such a big deal, except it didn't manifest itself until Maven ( m2e ) broke and through up a few thousand errors, none of which were as simple as "JDK missing".  This little gaff cost me two hours of frustration and I am familiar with Eclipse and Java!

I can't help but think if Google embraced NetBeans instead of Eclipse things would be much much much friendlier to new developers.  Then that seems to be a fault with Google in this first place, they treat everyone like enterprise developers, I saw it in Android, I saw it in App Engine and GDT and now I see it in PlayN.  

EDIT: Is this the right place for the guide, or should I have posted under tutorials?

[st33d]

I think a repost in Tutorials of your guide would be welcome

I also think we've pretty much covered that PlayN is specifically for Java developers dealing with the issue of browsers not supporting Java anymore.

If I was working as a Java dev right now, then I'd welcome PlayN, but they won't get any converts from Flash or Javascript. But to be honest, I don't think they need any.

[Chromanoid]

I can't help but think if Google embraced NetBeans instead of Eclipse things would be much much much friendlier to new developers.

Definetly. I prefer Netbeans too. It's a pity that GWT4NB seems to be dead -.-.

[mcc]

Yeah, truth of the matter is, someone exposed to Java development ( and Eclipse in particular ) will look at that guide and go, meh, that's not so bad.  For someone experienced with Java, Maven, Git and Eclipse it really is pretty typical, the process of working on Android is remarkably similar.

The problem is, Google are targeting this at HTML5 and Flash developers and in that case, learning all those other technologies just adds to the learning curve.  They are really going to have to streamline the process, like both of you said, if they want to appeal to non-Java developers.

At the core though, I think a lot of the complexity is attributed to Eclipse.  I know there are a core of people that absolutely love it, but I generally find it a pain in the ass.  When I was working on Android last year the environment corrupted itself so many times I lost count.  Then even when working on this tutorial, I started from a scratch install and Eclipse latched itself to the wrong java version ( the JRE instead of the JDK ).  This isn't such a big deal, except it didn't manifest itself until Maven ( m2e ) broke and through up a few thousand errors, none of which were as simple as "JDK missing".  This little gaff cost me two hours of frustration and I am familiar with Eclipse and Java!

I can't help but think if Google embraced NetBeans instead of Eclipse things would be much much much friendlier to new developers.  Then that seems to be a fault with Google in this first place, they treat everyone like enterprise developers, I saw it in Android, I saw it in App Engine and GDT and now I see it in PlayN.  

EDIT: Is this the right place for the guide, or should I have posted under tutorials?

What I find fascinating is comparing Adobe Flash Builder to the Android tools. Both are based on Eclipse. But in my experience with both is that FB is lovely to use, whereas the Android tools are a constant source of frustration that take forever to set up and need constant fiddling just to keep working day to day. (Also FB costs like $250 whereas the Android tools are free.) This makes me think Eclipse is maybe not the problem.

If the PlayN eclipse tools are as easy to use as the Android stuff I used...

I also think we've pretty much covered that PlayN is specifically for Java developers dealing with the issue of browsers not supporting Java anymore.

That is useful! I am very interested in PlayN for this exact reason!

[Chromanoid]

I think eclipse is a bit like a desktop. It gets easily bloated with unnescessary icons and shortcuts. Vanilla java eclipse is nice to use. I don't know why, but Netbeans can do this better.

Did anybody try to target Flash with PlayN yet? Is it possible to target iOS with PlayN's Flash output?

[Serapth]

I wonder how long until NetBeans is made terrible by Oracle.  I don't know if you ever experienced any of the Oracle database tools but... they make Eclipse seem like a gift from the gods.


Eclipse has potential and a whole lot of power, but you are right the extensions really seem to screw it up.  Then again, that's what happens when you have an everything and the kitchen sink approach. I am surprised Microsoft has managed to keep Visual Studio so cohesive with all the crap embedded in it now. As to the earlier post, I do wonder how much is Google's fault and how much is Eclipse's.  Frankly the only times in recent memory I have used Eclipse, it's been working with Google technology.

I do recall however, one of my first jobs out of school was working for a bank preparing to port their OS/X C++ code to Java and they used IBM Visual Age C++.  It was SLOOOOOWWWWW, horrifically slow, but amazingly powerful, did things other IDE's of the day could only dream of.  I think in the end the Visual Age product line actually became the genesis of Eclipse.

I feel old now.

[Chromanoid]

Well Oracle has JDeveloper, I don't think they want to change much in Netbeans. I hope Netbeans gets enough support from Oracle. The projects in the Netbeans Platform Showcase give me hope that they will continue to do so.  

[Serapth]

A bit of a necro, but to those interested in checking out PlayN, but put off by the install process, I just created an installer details here in the TigForums tutorial section


It is a really cool indie tech, wrapped in a really stupid install process.  If you are thinking about targeting HTML5, Android, Flash or the desktop and are a fan of Java, I highly recommend you check it out.  It's actually quite nice once you take out all of the install headaches!

[Player 3]

Ah, so it's pretty much an exporting library for Java, right?

[Serapth]

Nope.  It basically automates the various installs and configurations you would have to do.

It installs/configures: JDK, Git, Android SDk and Maven, run's Git to download the latest sources, runs maven on sources, then pops up a wizard for your new game, then crafts ( and runs ) the corresponding maven command to create your game.

Essentially it allows users to completely ignore the existence of Git, Maven and the Android SDK and get down to writing code. Well, other than the fact you use Maven to run your projects later...

Best of all, it takes away all the (super fragile) Eclipse dependencies, because frankly, they suck.  I am no fan of Eclipse, and this process allows me to use IDEA or NetBeans 100%.

[Chromanoid]

Wonderful! I will check it ASAP... (which will be in a month -.-)

[gnat]

I do not trust Google to keep this alive after reading this: http://googleblog.blogspot.com/2011/10/fall-sweep.html

The word is they shutting down Google Code Search next year. I encourage everyone here to send Google an email... maybe they'll decide to keep it with enough pressure from the programming community.

[Klaim]

I don't know. I understand why they are shutting down code search, it's rarely as helpful as just searching in normal google some code or library names. They certainly have statistics to decide to shut it down.

So the idea here would be more that if playn isn't used much (as I think it will be) they will shut it down. That said, it's a library, not a service. Stopping working on it will not kill it if there are still users.


Until they shut down Proejct Hosting 

[bateleur]

Eclipse has potential and a whole lot of power, but you are right the extensions really seem to screw it up.  Then again, that's what happens when you have an everything and the kitchen sink approach.

Only because they do it wrong.

Way back in the 90s I remember being impressed when Tcl/TK finally solved this problem. Want a new widget? You just declare one and it appears, in one very short line. All the vast complexity which some widgets support is hidden from the developer unless they want it. Compare with X11, where even "create simple window" takes nine parameters!

This simple principle seems to have been forgotten: make everything optional.

[Serapth]

Eclipse has potential and a whole lot of power, but you are right the extensions really seem to screw it up.  Then again, that's what happens when you have an everything and the kitchen sink approach.

Only because they do it wrong.

Way back in the 90s I remember being impressed when Tcl/TK finally solved this problem. Want a new widget? You just declare one and it appears, in one very short line. All the vast complexity which some widgets support is hidden from the developer unless they want it. Compare with X11, where even "create simple window" takes nine parameters!

This simple principle seems to have been forgotten: make everything optional.

Qt was such an amazingly clean design.  I think what hindered it most where

a) it was kinda ugly
b) it's initial license sucked.  Free on Linux but commercial on Windows?  And there was ambiguity about its viability in open source, much like Mono faced recently.


Both of these problems were eventually reconciled, but the damage was done.



Oh, and for the record, to those not monitoring the Tutorial thread... they entered PlayN in Maven Central which makes the install a hell of a lot easier.  Plus it's much easier to use Netbeans now, which is A Very Good Thing (tm)


Oh and on Eclipse, even though I hate it, I actually believe it was evolved from one of my favorite IDE's that nobody used, VisualAge.  I used Visual Age C++ for OS/2 in a prior job, and aside from the fact you needed a super computer to run it, it was simply amazing compared to it's peers of the day.

In seeing a screenshot of it, I am starting to feel extremely old: