New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
fix tachyon segfault introduced by #6542 #7232
Comments
comment:1
|
This comment has been minimized.
This comment has been minimized.
comment:3
(Still) Reproducible with Sage 5.8.beta3... :-) |
comment:4
Interestingly, it segfaults after having read the whole input file (and also with just one thread): (sage-sh) $ tachyon tachyon-example.dat +V -numthreads 1 -format PNG -o tachyon-example.png
Tachyon Parallel/Multiprocessor Ray Tracer Version 0.98.9
Copyright 1994-2010, John E. Stone <john.stone@gmail.com>
------------------------------------------------------------
Scene Parsing Time: 0.0052 seconds
CPU Information:
Node 0: 2 CPUs, CPU Speed 1.00, Node Speed 2.00 Name: sleepless
Total CPUs: 2
Total Speed: 2.000000
Scene contains 514 objects.
Global bounds: -0.1 -1.1 -1.1 -> 0.1 1.1 1.1
Creating top level grid: X:12 Y:12 Z:12
Grid: X: 12 Y: 12 Z: 12 Cells: 1728 Obj: 513 Obj/Cell: 0.297
Scene contains 1 non-gridded objects
Allocating Image Buffer.
Preprocessing Time: 0.0088 seconds
Segmentation fault: 0% complete (But the segfault actually completes. ;-) ) |
Attachment: tachyon-example.dat.gz Tachyon input file for the example given in the ticket's description. (Plain text) |
Attachment: tachyon-doctest.490.dat.gz Tachyon input file generated by the example in |
comment:5
Tachyon seems to bomb out on the original example in the limit as camera_center->(0,0,0). Then, from tachyon.py viewdir=(0,0,0), which may not make sense; although on other examples I've tried when camera_center and viewdir are the zero vector then Tachyon returns an empty image. For the given example using
does give, a not very interesting, image. How small vector_center can be and not cause Tachyon to bomb out my be hardware/software specific. With
I get an image similar to the parametric_plot() output. This sounds like a Tachyon bug. |
comment:6
Just built Sage 5.8.rc0 on a Linux x86 box (Pentium4 Prescott, GCC 4.7.2), and there testing
(One only sees this with |
comment:7
Replying to @nexttime:
Ooops, perhaps more importantly, verbosely testing |
comment:8
... after 75+ minutes CPU time, this Tachyon is still at 0% rendering progress for the example from the description ... 8-/ |
comment:9
... while the example from the doctest renders quick and well. |
Changed keywords from none to tachyon |
Commit: |
Branch: u/chapoton/7232 |
Author: Frédéric Chapoton |
Branch pushed to git repo; I updated commit sha1. New commits:
|
Branch pushed to git repo; I updated commit sha1. New commits:
|
comment:20
I think this is a reasonably way to handle this given how long it's been, and a good doctest, but a brief reminder here about what the default camera center and look at are might be helpful. |
comment:21
Replying to @kcrisman:
Actually, I think I like the error message better the way it is. Listing the defaults will be hard to maintain (in the unlikely case that they change later), unless we print them from some attribute. But that makes printing the error depend on successfully determining the value of the attribute, making it slower and more prone to error itself. I think the error message currently gives all the necessary information with a minimum of overhead and complication. Maybe as a compromise, if necessary, print "see init method for default values" ? |
comment:22
Fair enough, okay.
Well, it's actually in the |
Reviewer: Karl-Dieter Crisman, Niles Johnson |
Changed branch from u/chapoton/7232 to |
This pretty much says it all:
Note, that this was I think caused by #6542.
CC: @sagetrac-mhampton @nilesjohnson
Component: graphics
Keywords: tachyon
Author: Frédéric Chapoton
Branch/Commit:
f8cb975
Reviewer: Karl-Dieter Crisman, Niles Johnson
Issue created by migration from https://trac.sagemath.org/ticket/7232
The text was updated successfully, but these errors were encountered: