site_archiver@lists.apple.com Delivered-To: pro-apps-dev@lists.apple.com Hi, Pauli, - Paul On May 23, 2008, at 9:24 AM, Pauli Ojala wrote: Paul, Best regards, Pauli -- Pauli Ojala Lacquer oy/ltd pauli@lacquer.fi On 14.5.2008, at 21.14, Paul Schneider wrote: Hi, Pauli, Thanks, - Paul <FloatingWindow.zip> On May 14, 2008, at 8:18 AM, Pauli Ojala wrote: Pauli Ojala _______________________________________________ Do not post admin requests to the list. They will be ignored. Pro-apps-dev mailing list (Pro-apps-dev@lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/pro-apps-dev/pschneider%40apple.com _______________________________________________ Do not post admin requests to the list. They will be ignored. Pro-apps-dev mailing list (Pro-apps-dev@lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/pro-apps-dev/site_archiver%40lists.ap... thank you very much for providing the FloatingWindow FxPlug sample. I've had the users that were reporting problems with our plugin in FCP 6.0.3 run the sample, and it works as expected for them, so the problem is clearly in our FxPlug code. Were you able to reproduce the problem with my bars & tone steps? My guess is that this what your users are hitting. Even if not, someone is bound to hit it eventually. Can you elucidate a bit as to why the getAPI call is necessary here, and why one must check for its return value before proceeding? I'm only getting the current value of the parameter so I can increment it to get the new value. Getting the value first shouldn't be necessary. You do have to make sure you are assigning a new value to the parameter (one that is not identical to the old value). Also, I notice you're releasing the value returned from the get method. Is it always the case that -getCustomParameterValue retains the object before returning? That doesn't seem intuitive (I had expected that the setter retains but the getter doesn't, like standard Cocoa classes). You're right - that was a mistake on my part. Values returned from - getCustomParameterValue shouldn't be released. It looks like FCP is incorrectly returning retained references from - getCustomParameterValue; I've filed a bug. Motion does NOT return a retained reference from getCustomParameterValue. thank you very much for providing the FloatingWindow FxPlug sample. I've had the users that were reporting problems with our plugin in FCP 6.0.3 run the sample, and it works as expected for them, so the problem is clearly in our FxPlug code. The only significant difference is in the -forceRerender method. I have a few questions concerning your implementation, more specifically this section: ---- NSNumber* oldValue = nil; if ([getApi getCustomParameterValue:&oldValue fromParm:kParamID_Button]) { NSNumber* newValue = [NSNumber numberWithInt:[oldValue intValue] + 1]; [actApi startAction:self]; [setApi setCustomParameterValue:newValue toParm:kParamID_Button]; [actApi endAction:self]; } [oldValue release]; ---- My old version didn't call -getCustomParameterValue at all, it just went straight ahead with the actionAPI/setAPI calls. I guess that's probably why my code started failing. Can you elucidate a bit as to why the getAPI call is necessary here, and why one must check for its return value before proceeding? Also, I notice you're releasing the value returned from the get method. Is it always the case that -getCustomParameterValue retains the object before returning? That doesn't seem intuitive (I had expected that the setter retains but the getter doesn't, like standard Cocoa classes). The FxPlug API reference doesn't seem to make any mention of the retain/release policy for these methods. If it's right that - getCustomParameterValue always retains, my code has been leaking a scary amount of parameter value objects... I'd encourage you to add a big warning about this in the documentation! there were very few FxPlug-related changes in Final Cut 6.0.3, and I'm not aware of any that might have caused a difference in our behavior here. I wrote a very simple FxPlug filter that manages its own floating Cocoa window, with a button in the window to trigger a redisplay. It seems to work correctly in 6.0.3. I've attached it - if it works on your systems, perhaps you can spot a difference between this plugin and what you're doing? Also, when you say "we're receiving reports", do you mean that you've tested your plugin in 6.0.3, and something that always used to work now always fails? Or do you just mean that some of your customers are reporting problems? Hello, I'm hoping someone on the FCP FxPlug team might be able to give some insight into a problem that our users have been experiencing. Our plugin Conduit uses a private Cocoa window for its nodal effect editor UI. When the user modifies something in the editor window, we use FxCustomParameterActionAPI and FxParameterSettingAPI to set a new value for a custom parameter on the FxPlug instance. Ever since Motion 2.0, this method has worked fine for triggering a render in the host application. This way we don't need to run the editor modally; the user can keep the window open while moving around on the host app's timeline, which is very convenient. Now we're receiving reports that the FCP 6.0.3 update has broken this behaviour. Apparently the Viewer doesn't update anymore in response to events in our editor window. However, one user mentioned that switching from FCP to Finder and then back to FCP does trigger the expected render in the Viewer in FCP. So it sounds like the problem might be some fairly trivial bug somewhere -- as in the viewer not getting the equivalent of [-NSView setNeedsDisplay:] after the FxPlug parameter has been modified...? This email sent to pschneider@apple.com This email sent to site_archiver@lists.apple.com