Ok thanks. Unfortunately both simple examples I tried didn't repro the problem. It might be that the garbage collector has to run for the problem to happen. I only experience the issue when a lot of work has been done between creating and using the list.
On 5/4/10, Thomas Paviot <tpav...@gmail.com> wrote: > 2010/5/4 Dave Cowden <dave.cow...@gmail.com> > >> Hey, all: >> >> I have had a strange problem occur a couple of times, wondering if anyone >> recognizes it. >> >> The behavior occurs when looping over shapes and dealing with them in >> python lists. During the iteration, >> all of the objects appear normal and correct. But later when they are >> accessed in the list, all of the references in the list appear to be the >> last object written into the list. >> >> >> For example, I had code like the below, which was transforming a list of >> shapes: >> >> wireList = []; #this is a list of wires that are all ok >> >> >> #make transformation >> p1 = gp.gp_Pnt(0,0,0); >> p2 = gp.gp_Pnt(0,0,z); >> xform = gp.gp_Trsf(); >> xform.SetTranslation(p1,p2); >> bt = BRepBuilderAPI.BRepBuilderAPI_Transform(xform); >> >> newList = []; >> >> for w in wireList: >> bt.Perform(w,True); >> newList.append(bt.Shape() ); >> TestDisplay.display.showShape(bt.Shape() ); >> >> >> >> When this code runs, the shapes displayed are correct. But later when they >> are accessed, what i get is a bunch of copies of the last one added. >> >> >> I have also seen this behavior before iterating over wires using a >> wireExplorer-- in that case, it was resolved by using the super-ugly and >> annoying TopTools.TopTools_HSequenceOfShape() object instead of a python >> list. >> >> Should I use HSequnceOfShape instead of python lists, or is there a trick >> that I can use to let me use lists? >> >> thanks! >> >> > > Hi Dave, > > I have to reproduce this issue from a simple example, and create a unittest. > > In the meantime, you may have to use the HSequnceOfShape. Although it's not > pythonic, it's more safe to use builtin OCC list manager rather than python > list. The mapping between python lists and OCC lists have to be performed at > the SWIG level. > > Best Regards, > > Thomas > _______________________________________________ Pythonocc-users mailing list Pythonocc-users@gna.org https://mail.gna.org/listinfo/pythonocc-users
