Why WebStart and not Apllet?

[m77]

I'm lame in WebStart. So what are adventages?

[Eli Delventhal]

Speed. Control. Non-reliance on a web browser.

[noblemaster]

Why not setup your project, so you can deploy both as Java Webstart + Java Applet.

Games published by our own members! Check 'em out!

[m77]

Speed. Control. Non-reliance on a web browser.

Tell me more about speed.

[fletchergames]

When you use WebStart, your program runs as an application rather than an applet.

I don't know all the details, but applets are going to be slower because there's an extra layer: the browser.  Also, I believe you can take advantage of the hardware better when you put your application in full-screen exclusive mode, which you can't do with an applet.  There may be other speed benefits as well.

[ewjordan]

I don't know about a tangible speed difference between WebStart and applets, but stability is a big issue: applets crash all the time and they hang browsers (if not permanently, at least while the applet is downloading, which can take a while).  Of course, WebStart is kind of annoying to a lot of users, and makes them click a security box.  Also WebStart breaks you out of the browser, which might mess with an ad-supported revenue model - perhaps if some ad network starts allowing in-game Java ads this will change, but I fear the market is a bit too small for that to be much of a priority.

[bienator]

there is no noticeable speed difference and with java 6 update 10 applets will not freeze any more the browser since they run in an separate process. (this was not only java's problem remember starting acrobat reader in a tab? The chances where high that you crash the browser if you've done that on low bandwidth machines)

Ken's J1 Session is a must:
http://news.zdnet.com/2422-13568_22-200560.html

[Eli Delventhal]

So basically it seems that the only real detriment is that you can't use full screen mode, which can really speed up rendering and free up more RAM for your game. Most indie projects probably won't care about this, however.

[CaptainJester]

I don't know about a tangible speed difference between WebStart and applets, but stability is a big issue: applets crash all the time and they hang browsers (if not permanently, at least while the applet is downloading, which can take a while).  Of course, WebStart is kind of annoying to a lot of users, and makes them click a security box.  Also WebStart breaks you out of the browser, which might mess with an ad-supported revenue model - perhaps if some ad network starts allowing in-game Java ads this will change, but I fear the market is a bit too small for that to be much of a priority.

You only get the security box for a signed application in WebStart.  It is the same with applets.

[gouessej]

So basically it seems that the only real detriment is that you can't use full screen mode, which can really speed up rendering and free up more RAM for your game. Most indie projects probably won't care about this, however.

That's wrong. Exclusive fullscreen mode doesn't speed up rendering significantly, only a very little bit. I did the test with TUER under Microsoft Windows NT 2000 and there was no difference of performance, the same under Vista.

Games published by our own members! Check 'em out!

[trembovetski]

That's wrong. Exclusive fullscreen mode doesn't speed up rendering significantly, only a very little bit. I did the test with TUER under Microsoft Windows NT 2000 and there was no difference of performance, the same under Vista.

That depends.. In 6uX (where X < 10) on windows in full screen mode Direct3D may be used for
rendering (as opposed to DirectX, which is the default). It won't make much of a difference
if the D3D pipeline couldn't be enabled.

In 6u10 if the new D3D pipeline can be enabled at all it is enabled in both windowed
and full screen modes. If it is enabled, you may see some difference in fs mode
just because there are less restrictions on what can be done in FS.

Dmitri
Java2D Team

[cylab]

That's wrong. Exclusive fullscreen mode doesn't speed up rendering significantly, only a very little bit. I did the test with TUER under Microsoft Windows NT 2000 and there was no difference of performance, the same under Vista.

With FSEM you afaik get the chance to choose your resolution and rendering on a lower res than 1920x1200 will surely speed up things