New j4k contest anytime soon?

[Markus_Persson]

Yes, allowing both would be better than excluding one.

And I still have faith in applets. ;D

[nonnus29]

I’m really surprised to hear that your applet is smaller. For an applet you have to give it a thread to run in and all that instead of using the constructor ???

I’m disappointed pack200 jars won’t execute, is this a fact or conjecture?

What are the complications with jws? I’ve never worked with it so I don’t know.

Applets are the no brainer solution, but I think having an ‘applet 4k’ contest is much different than a ‘java 4k’ contest. For one you’re adding a restriction that is new and for another, like it or not, applets have a bad reputation. I for one like compileing without deprecation warnings… Note, I think I had the only applet last time.

[quote]And I’m not going to enter if an executable jar file is required.
[/quote]
Cool, now there’s more room in the top ten!!!

;D

[Alan_W]

I prefer an executable jar myself, but wouldn’t preclude applets.

If pack200 only works with webstart, it would be difficult for entrants with only their ISP provided webspace, since these often don’t have the required mime types. In addition, unless signed, webstart entries cannot go full screen or set the ‘nodraw’ property. Also when running (unsigned) in a window, the ‘Java Application Window’ banner grabs some of the bottom of the main window. Allowing pack200, would put executable jars and applets at a size disadvantage. Overall it’s a mixed blessing. I’d prefer not to allow it really, although thats the Luddite in me talking.

I’m still tempted to code for java 1.4.2, although 1.5 should be allowed. There are probably still quite a few computers out there without 1.5.

Alan

[hvor2]

@jbanes

Webstart. Now THAT is sexy. Applets, no. Of course, everyone is entitled to their own opinions, but mine is that Applets should not be pushed.

This is my opinion, too. Aplets can’t be compared to webstart possibilities.

[darkprophet]

Why not just say use webstart and the JNLP is not included in the size?

One would then argue that you can pass entire data throught it as arguments, but just forbid arguments…

This way, applets, applications, and any other thing that starts with app will work

DP

[jbanes]

@darkprophet

That’s what we’ve been doing. The problem with Webstart, however, is that you need to sign the JAR to do anything useful. For example, if you want to switch to full-screen mode, your app MUST be signed. The signature itself is a rather nasty size penalty, too.

The other consideration is that Webstart makes it more difficult to obtain a permanent copy for testing. At least with Applets you can give the webpage and be happy. The codebase requirements for Webstart make that just a little trickier.

In short, reducing entrants to only executable JARs or Applet form makes judging easier, makes user’s lives easier, and even makes the lives of contestants easier. (Even if they don’t realize it.)

[quote]If pack200 only works with webstart, it would be difficult for entrants with only their ISP provided webspace

Just to be clear, Sun does say that you can use it as an archive format similar to ZIP. It’s only technically “supported” by webstart because webstart automatically unpacks it into its cache.

[quote]I’m still tempted to code for java 1.4.2, although 1.5 should be allowed.
[/quote]
Setting the rules at 1.5 shouldn’t preclude anyone from targetting 1.4. All it means is that your work will be judged under a 1.5 VM. Since Java is so good about backward compatibility, that shouldn’t be a problem for most.

[nonnus29]

Since pack200 gives such a tremendous compression advantage and can only be used for webstart it should be excluded from the competition entirely.

Or…

Allow it and for those who really care about the extra 500+ bytes and want to go thru the trouble of signing jars, setting up jnlp, and allowing only those with jdk 1.5 to play there game then they can.

What do you guys think?

[Alan_W]

Has anyone tried out pack200 on an existing 4k game? I found this quote:

I had a chance to check out the spec. Basically, the algoithm takes into account strings pools of all classes being packed and package them as a one cental pool. Executable bytecode is not being compressed.

Since there usually is only one class in a 4k program & little text, there may not be a big advantage in using pack200.

NB. IIRC, the main drawback of the zip format is that it compresses each class separately, which is not really an issue if you only have one class.

Alan

/Edit: Found this old post by Abuse on sun forums. Kinda bears out my suspicions. I am beginning to think that pack200 is a red herring.

[c_lilian]

@nonnus29

Since pack200 gives such a tremendous compression advantage and can only be used for webstart it should be excluded from the competition entirely.

pack200 works also for applets…

Lilian

[nonnus29]

@Alan_W

Has anyone tried out pack200 on an existing 4k game? I found this quote:

I had a chance to check out the spec. Basically, the algoithm takes into account strings pools of all classes being packed and package them as a one cental pool. Executable bytecode is not being compressed.

Since there usually is only one class in a 4k program & little text, there may not be a big advantage in using pack200.

NB. IIRC, the main drawback of the zip format is that it compresses each class separately, which is not really an issue if you only have one class.

Alan

/Edit: Found this old post by Abuse on sun forums. Kinda bears out my suspicions. I am beginning to think that pack200 is a red herring.

Nope, pack200 works super well, and it works for applets like C. said.

My test:

jar: 4.52kb
pack: 3.03kb

and it runs as an applet with an ‘archive’ property in the applet tag. That’s ALOT of uncompressed code!

Yes, I’m using applets from now on ;D .

[Markus_Persson]

Holy moly!

[Morre]

I might be wrong, but as far as I can recall, the rules last year allowed basically anything, as long as one could start it in one way or another. This seems like both a good and a bad idea.

Last year, for webstart files, you were allowed to exceed 4k with the signed JAR, right? And the JNLP was not included in file size. That’s good, I think.

I can’t claim to know anything about Pack200, but if it gives that much of a size advantage, it really uppers the practical limit a lot, allowing for much bigger games. Being a 4k contest, I don’t really see the point, so I think this should be disallowed.

I’d personally prefer if you could use either self-excutable JARs or Applets. Furthermore, I’d allow webstart, excluding the signature’s extra size, and the JNLP file, by forcing anyone using webstart to create a self-executable JAR that does exactly the same thing (in other words, no specific rules for webstart - let anyone who wants webstart follow the normal rules, and then create another file for webstart).

If those were the rules, I’d surely post any entries as JARs and webstartable (JAR for the contest, webstart for opinions, accesibility, etc).

[Alan_W]

@nonnus29

Nope, pack200 works super well, and it works for applets like C. said.

My test:

jar: 4.52kb
pack: 3.03kb

and it runs as an applet with an ‘archive’ property in the applet tag. That’s ALOT of uncompressed code!

Yes, I’m using applets from now on ;D .

Interesting. Was the starting jar already obfusticated & compressed with 7zip (compression=9) or kzip? It’s important for comparison, since I usually manage to shrink the stock jar by nearly 50% that way.

Alan

[jbanes]

@Morre

Incorrect. Many developers offered a Webstart version for convenience. However, most were actually judged on a downloadable JAR version of the game.

[quote]And the JNLP was not included in file size.

Correct.

[Markus_Persson]

[quote=“Morre,post:52,topic:25185”]
I disagree strongly. The point of the competition is to try to squeeze as much as possible into 4k. Preventing means of doing so is counter-productive and silly.
If you don’t mind me doing a bit of slippery-sloping, if we ban pack200, we should also ban jars with non-zero compression.

[Alan_W]

I pack200’d my provisional entry:

frag4k.jar = 4094 bytes (obfusticated with proguard & zipped with 7zip at max compression)
frag4k.jar.pack.gz = 3262 bytes (now 3246 bytes with added secret sauce)

Ok, I’m convinced there’s a saving.

Alan.

[mlk]

We should start a new one.
I’ve been completed out of the loop. I’ll try to catch up on this thread, and contact Woogley.

[Morre]

Jbanes: I think you misunderstood me, that’s just what I meant… about the webstart, I mean - the signed jars aren’t specifically allowed, but rather something you can do if you want to, IF you also have a normal jar that’s less than 4k.

Markus_Persson: I see your point - either packed, or not packed. If packed, pack it all the way… I guess you’re right, although it does alter the contest possibilities in a rather extreme way, doesn’t it?

Sorry mlk, I’m done now

[nonnus29]

To clarify:

You can pack200: applets and jnlp

You can pack200 an executable jar file but it has to be unpacked before it can be executed. Applets and jws do this for you.

[kevglass]

Why not get something fun and and original working before worrying about the 4k limit, Not to say that anything I’ve ever produced is original but the shrinking things down after the fact reallt isn’t that difficult.

Kev