Going to release?

[Jeff]

Hey Guys.

I want to make an architectural change.  Its minor but it will break the plug-ins, so I was thinking maybe we should lock down a 1.0 version before I do that.  Everyone think we're there?

The change i want to make is add one more required method to plugins: isSupported().  This way a plug-in on initialization can decide it doesnt belong in this environment.  (eg the Win32 plug-in being used on a Mac) and just set isSupported() to return false.
The main API will then check this and throw out any plug-ins that don't return true so the user never sees them.

Sound reasonable?

JK

[endolf]

Hi
 There is a bug in the linux stuff that I wouldn't mind fixing first, but it depends how important it is (no-one has reported it yet). My logitech wingman cordless rumbling gamepad get's picked up by the joystick driver as having 7 axes (when it only has 5, but 2 rumblers too), and the event driver picks it up with 5, then in my code I see that it's a different axes count and get confused. Supposedly you can get these working in the kernel, but I've failed so far. I don't know wether any other force feedback stuff has the same problems because I don't have any. What I was hoping to do was get the rumbler support working on the linux plugin and in the process solve this, but other projects have taken priority as no-one else has reported it.
 Let me know if you want this (or the fix) in before 1.0, but as far as I can see it can probably wait till after.

Endolf

[gregorypierce]

Someone send me an email or something that outlines the architectural changes so I can make sure the OSX stuff works properly.

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

[Jeff]

Okay.  We wont do it I think til we settle down again from the changes Endolf is making now (rumbelr support and hopefully final JAWS support.)

I've been thinking about it and I MAY actually be able to hide the change from the old plug-ins.  (If they all inherit from a super-class I can put a dummy test in the superclass to be over-ridden.)

I'll try to look at it sometime soon.

JK

[endolf]

Okay.  We wont do it I think til we settle down again from the changes Endolf is making now (rumbelr support and hopefully final JAWS support.)

D'oh!, I was hopeing the rumbler one didn't get out, incase I fail miserably

I've been thinking about it and I MAY actually be able to hide the change from the old plug-ins.  (If they all inherit from a super-class I can put a dummy test in the superclass to be over-ridden.)

Although not commited (cause it's all broken on my cvs node right now ) both the windows and linux one have the isSupported() method (nothing calls it though, i'm leaving that out on purpose), it just checks the os property just like ant does before building
 Had another though, if the property jinput.plugins is set, should the isSupported method be checked?, the jinput.plugins could be seen as a 'force this plugin to load' kinda option too, say for someone who is running a version of linux with a funny os string, or on some VM that didn't set the property properly?

Cheers

Endolf