-
O.K. As a bit of a test, I have committed the two patches
I posted on New Year's Day, and also started the NEWS
file.
I tagged the HEAD before doing anything, but the project
in its former state could in any case be retrieved using
a date of 2004-12-31.
Another problem is that there are files called Test.cpp
and test.cpp in the same directory, and both are
used.
I will move over...
2005-01-18 18:46:18 UTC by bpfowler
-
bpfowler committed patchset 30 of module templates to the zRcube CVS repository, changing 2 files.
2005-01-18 17:42:14 UTC by bpfowler
-
Hi
I'm back from skiing =). I just added your user to the project so you should be able to commit your patches on the CVS.
Be careful with the povparser : bison version are often not compatible together (one of the bison maintainer was one of our teacher and we often used 'experimental' versions). Moreover, we used a bison++ (the c++ version of bison was not available at this time) which...
2005-01-04 17:47:30 UTC by mandor-
-
>I am still working on getting a Linux system that will
>run gcc-2.95 .
>
>I will have a play around over Christmas and send you
>some patches in the New Year.
I am sitting at a G4 running an up-to-date version
of Gentoo, which has a slot for gcc 2.95.3 and a
gcc-config script both of which still work. With gcc
set to 2.95, zrcube compiles as is (obviously I happen...
2005-01-01 13:11:41 UTC by bpfowler
-
> I have written a fairly long reply to the various parts
of this. Would you like me to post it?
Why not ? You can send it by e-mail if you prefers : mandor (at) lrde.epita.fr
About the suggested improvements, I agree. Of course.
--
Mandor.
2004-12-23 15:15:05 UTC by mandor-
-
> We learnt a lot with this project and ...
I have written a fairly long reply to the various parts
of this. Would you like me to post it?
I am still working on getting a Linux system that will
run gcc-2.95 .
I will have a play around over Christmas and send you
some patches in the New Year.
As I mentioned earlier, I suspect that automake/autoconf
needs updating - committing a...
2004-12-23 04:02:12 UTC by bpfowler
-
It is not a one pixel shift. I was thinking that there were
some random numbers (Mone Carlo), but the report
didn't mention calls to 'Noise'.
I was comparing the pictures to see if there were any
problems with the build. If my build were O.K., I would
expect the two to be identical.
2004-12-22 15:39:39 UTC by bpfowler
-
Another possible explanation : the example picture could have been computed using the smp (multi-thread) or the "distributed" code. This code uses some arithmetics to distribute the pixels to render. It's not impossible that a small bug in this arithmetics causes a one-pixel shift somewhere (for example at the beginning of each line, ...) . Or maybe there's a real bug somewhere.
Btw,
2004-12-22 14:36:59 UTC by mandor-
-
> Did you check that the generated pictures have the same
> size (width/height of the picture) ?
There is absolutely no visual difference and GQview shows
99% duplication.
However, if I convert both images to ppm and pipe through
hexdump, I can show that every pixel differs by 3 to 5 pixel
values (in all three channels) but irregularly.
2004-12-22 10:22:37 UTC by bpfowler
-
> Did you check that the generated pictures have the
> same size (width/height of the picture) ?
Both are 512 by 384.
I assume that I am missing some options, Jaguar
does not come with getopt_long
http://sourceforge.net/mailarchive/forum.php?thread_id=3926732&forum_id=3248
(but Darwin does...
2004-12-22 05:29:37 UTC by bpfowler
