IOhannes, you are right... I try open one more time testing-abstration (sorry for the error..) in windows that time. And outlet is showned in autotips, the only problem it's a little hard to hit the outlet area...
If can be fixed the problem of rectangle GOP and IOlets always at the top, I'm a very happy pd-user! ;)
I can suggest either a "little bigger" area of autotips response to iolets.
2013/2/18 Jonathan Wilkes jancsika@yahoo.com
From: Esteban Viveros emviveros@gmail.com To: Roman Haefeli reduzent@gmail.com Cc: PD-list list pd-list@iem.at Sent: Monday, February 18, 2013 10:18 AM Subject: Re: [PD] GUI abstraction issues
Well I try the IOhannes sugestion, and autotips don't work correctly with
my abstraction... :/
I tried to create another example (testing-abstration), and now only the
inlet are working with autotips feature... In pd-l2ork everything works fine.
Indeed it doesn't. It looks like when the abstraction is GOP, it registers all the xlet #s as "-1", and it doesn't detect the bounding box of the object. (The latter because I'm looking for the text tag, which a gop box doesn't have.)
I imagine Ivica's works because he's checking for tooltips on the c-side, and I'm checking on the tcl side.
Pd-l2ork: Check whether there's an object or inlet under current mouse position. If there is, look for its tooltip text in the help patch, then draw the tooltip.
Pd-ext: Trigger an event when the mouse hovers over an object or inlet that is _visible_ on the canvas. Look for the tooltip text in the help patch, then draw the tooltip.
Pd-l2ork's approach is more robust-- for example, it's easier for him to enlarge the area that will register an xlet tooltip. Pd-extended's currently has what I think is the "proper" behavior, though-- you only show tooltips for objects that are visible to the user. If the patch author covers up an object with a [cnv], he/she is actively hiding that object, and it shouldn't register a tooltip. It's bad UI design to show a tooltip when the user can't see the thing it refers to.
Anyway, I'll investigate and see if I can come up with a fix for the testing-abstration (sic) situation. But if you're putting [cnv] objects on top of xlets then they shouldn't respond to tooltips. That problem should be fixed at the source: xlets and object rectangle should always be at the top of the stacking order. (I guess there could also be an option for hiding xlets or the box, but that doesn't seem like a good idea for visual dataflow.)
-Jonathan