This behavior comes from the pdtk_canvas_place_window proc in pdtk_canvas.tcl which, as others have noted, uses the declared width of the overall desktop aka the widths of the screens, but does not take into account their relative positioning. In digging further, it seems Tk provides ways to check the offset to the main screen via the winfo vroot* queries. I'll try some tests tomorrow and it's probably an easy fix...

On Sep 14, 2017, at 10:24 AM, pd-list-request@lists.iem.at wrote:

Date: Thu, 14 Sep 2017 10:23:38 +0200
From: Raphaël Ilias <phae.ilias@gmail.com>
To: Nicolas Montgermont <list@nimon.org>
Cc: pd-list <pd-list@lists.iem.at>
Subject: Re: [PD] trouble with pd-0.48-0 and multiple screens
Message-ID:
<CADYAdN-4rBd2J08D5abKp8H8XtooxNqshPtNo5vdx01_YRiYvQ@mail.gmail.com>
Content-Type: text/plain; charset="utf-8"

Roman, for me it happens when i try to open (again) a pd-subpatch that i
previously placed on the second monitor (on the left of the main monitor) :
instead of appearing in its previous position, the window appears outside
the screen (on the right of the right screen), so you can't see it, drag it
back...
Yes Nicolas, I experienced this with the "additional" monitor on the left,
so subpatches should probably have negative coordinates.

Hope this gives a clue to solve this, patching on two monitors is nice for
big patches...

all best

raphaël

--------
Dan Wilcox
@danomatika
danomatika.com
robotcowboy.com