Hi there!

I was trying to build a simple patch (arrayleakagesubpatch.pd attached) with some visible controllers, and an array that should only be visible when you open this patch. I am using Pd 0.39-2. I placed a bang within the red square defined in the graph-on-parent properties and the array outside this area. I assumed it would work just like other controllers (bangs, sliders), but apparently the information on the array leaks to the parent patch (see arrayleakage.pd attached), whereas other controllers (e.g. a bang) don't leak. I am not sure if this is a bug or a feature, so any comments are appreciated!

The only way I found to prevent this leakage was to remove the graph-on-parent property of the array (inside arrayleakagesubpatch.pd), but I really wanted to be able to see its contents when I open this file. I tried to push the array to a subsubpatch (so that the array would still be visible on arrayleakagesubpatch.pd), but it leaks to the parent anyway.

Any comments welcome! Thanks in advance!

Marcelo

http://www.pdpatchrepo.info/hurleur/arrayleakage.zip

hide the array inside a subpatch

http://www.pdpatchrepo.info/hurleur/arrayleakagesubpatch.pd

Thank you for your reply!

I had explored this alternative, and although it solved the leakage problem,
it didn't solve my other problem, which was, I would still like to open this arrayleakagesubpatch and see the contents of the array without having to open another window.

If the first version's leakage is not actually a bug, then I probably misunderstood the function of the red sub-rectangle when you create graph-on-parent patches. This is quoted from the 0.39 release notes:

Graph-on-parent subpatches and abstractions no longer scale the GUI objects to fit the parent rectangle; instead you get a sub-rectangle in the subpatch, of the same size as the parent object, to place GUI objects in. GUI objects that don't fit inside aren't shown on the parent, and the parent objects no longer stretches itself to show things that wouldn't otherwise fit.

My point is: either that array should not appear on the parent (arrayleakage.pd) and this is a bug OR an array does not behave as either a GUI object or a regular object for the purpose of graph-on-parent subpatches, because those objects are supposed to be hidden on the parent. Is the latter the case? I wonder if this does not jeopardizes the whole idea of the red sub-rectangle.

Thanks again for your comments!

Indeed, your point is well taken, I also see it as bug rather than feature.
The red box is exactly as you say, to delimit the visible area.
Scaling was a brave attempt to get around the differences in how a patch might
look when rendered across platforms, fontsets and so on...but it relied on the size
of the window, it tried to push objects around so they would fit, oh man don't even
talk about those days... the red rectangle is the correct way, leaving placement to
the interface designer now that the view is much more uniform because a lot of
good work has been done on fonts, appearance, uniform pixel widths etc.

You can already hide an array because the array is invisible itself, what tou see
est a graph. If you turn off GOP in the array box then the graph it's hidden...
hmm not to kick off a discussion here that belongs on devlist but there's a few
quirks with arrays and graphs still, I think we all agree that this leakage is
undesirable but afaik the only fix now is to shove them in a subpatch.

regards,

a.

Thank you for your comments. I'll try and see if there is no bug report on the subject.

Regards.

hey,

i've been trying to find a way around this bug for ages, but nothing works really ...... using subpatches or covering the leakage with a white (or else) canvas after each loading are temporary solutions i use, but if you find anything else .....

This probably isn't ideal, either, but at least you don't have to open a new window to see the array. Basically, the array is in a subpatch with GOP being toggled on and of. When you hit the [bng] in the parent window, it first makes sure GOP is off before doing the calculations (using [t b b]), so it never tries to draw itself in the patch. When you open the abstraction, you have to click the [tgl] to see the array.

http://www.pdpatchrepo.info/hurleur/arrayleakagesubpatch.mmb.pd

hey maelstorm,

good one this works ! where did you get to know about donecanvasdialog? (i checked everywhere in the tutorials etc). are there more of these life savers?

just a questions:

• does a gui eat up cpu even when it is not showing? (like faders following automation etc...)

hey me again,

tried using that tool with my prblematic patch, and it seems to do the work, and it answers my above question: a gui eats up cpu only when its showing ...
thanks loads, i'll stick the patch when i complete it.

gab