Output of a two functions in sage.plot.plot3d.base is nondeterministic due to use of set

[roed314]

Graphics3d.tachyon and Graphics3d.mtl_str both join string representations of the textures in Graphics3d.texture_set(). The order of elements pulled out of texture_set() can be different.

CC: @jpflori

Component: graphics

Author: David Roe

Reviewer: Karl-Dieter Crisman

Merged: sage-5.0.beta11

Issue created by migration from https://trac.sagemath.org/ticket/12382

[roed314]

comment:1

Attachment: 12382.patch.gz

[kcrisman]

comment:3

I'm not quite sure why this happens - all is fine with or without on Mac, I know set is system-dependent? - but this definitely is a good fix and it's correct in the doctest. Can you tell me where it didn't work? I assume this was discovered due to a doctest failure, but it must have been on an uncommon platform, perhaps?

Anyway, modulo that, positive review.

[kcrisman]

Reviewer: Karl-Dieter Crisman

[roed314]

comment:4

It was discovered while developing the new doctesting framework at #12415.

[kcrisman]

comment:5

Ok, but I mean on what platform? I'm concerned to try to test that this works properly everywhere. Any other details would be helpful.

[roed314]

comment:6

It was on OS X 10.6, but using different doctesting code (that hasn't been posted to #12415 yet). Currently I don't know of any platforms where this problem arises. The objective of this ticket is to make #12415 less of a patch-bomb. :-)

[kcrisman]

comment:7

Ok... well, I guess since I tested on that platform I can say this is fine. But it seems odd to me that this would make a difference regarding which order the set happens in, if nothing else changed... Anyway, comparing pictures of

G = icosahedron(color='red') + sphere((0,0,0), .6, color='yellow')

in Tachyon before and after the change they seem the same, and everything else seems good.

[jdemeyer]

Merged: sage-5.0.beta11