Spec URL: http://www.auroralinux.org/people/spot/review/vamos.spec SRPM URL: http://www.auroralinux.org/people/spot/review/vamos-0.5.5-1.fc6.src.rpm Description: Vamos is an automotive simulation framework with an emphasis on thorough physical modeling and good C++ design. Vamos includes a real-time, first-person, 3D driving application. It also includes caelum, which is a program to help make seamless sky boxes from panoramic photos. The image file is mapped onto a sphere and six views are shown. The views are from the center of the sphere looking up (top row), down (bottom row), front, right, back, and left (middle row). When these images are mapped to the faces of a cube and viewed from the center of the cube, the view in any direction is indistinguishable from the view from the center of the sphere. The images for the sky box are extracted by taking a screenshot.
MUST FIX: * Please explain in detail, why you want to include a static lib or disable it (--disable-static). Feel strongly encouraged to do the latter. Some further remarks: 1. The configure script produces bogus results: ... checking for IceConnectionNumber in -lICE... no .. On FC5: nm -sD /usr/lib/libICE.so | grep IceConnectionNumber 00b1f530 T IceConnectionNumber 2. The *.info-install-rules are broken. The package doesn't ship/build *-<N>.info etc., just *.info.
Nuked the static lib, fixed the -lICE check, fixed the info-install rules. New SRPM: http://www.auroralinux.org/people/spot/review/vamos-0.5.5-2.fc6.src.rpm New SPEC: http://www.auroralinux.org/people/spot/review/vamos.spec
Odd that this has sat around for so long. I tried to build it in mock on x86_64 rawhide but wasn't successful. I know very little about GL stuff but I guess it needs build dependencies on mesa-libGLU-devel and freeglut-devel. Unfortunately even after that it fails to build. The output is a mess because of parallel make, so I disabled it and got this: Car.cc: In constructor 'Vamos_Body::Car_Reader::Car_Reader(std::string, std::string, Vamos_Body::Car*)': Car.cc:547: internal compiler error: output_operand: invalid expression as operand Please submit a full bug report, with preprocessed source if appropriate. See <URL:http://bugzilla.redhat.com/bugzilla> for instructions. {standard input}: Assembler messages: {standard input}:29088: Warning: end of file not at end of a line; newline inserted Preprocessed source stored into /tmp/ccULqHNF.out file, please attach this to your bugreport. Sure doesn't look good. I'm not really sure what to do at this point.
Let me revisit this package, it had fallen completely off my radar. I'll put up new SRPMS and SPEC when its ready to be reviewed.
Created attachment 157850 [details] Proposed spec changes I am proposing these changes to you spec.
Created attachment 157851 [details] Proposed diff This package's configuration is pretty broken. To work around them, I hacked its configuration by brute force and reran the autotools. The result is the patch inside of this attachment. With this patch and my patched spec applied the package build without problems on x86_64, i386 and ppc64. Probably still needing further fixes: *-devel's Requires.
OK, new version of vamos has a much cleaner configuration. I absorbed Ralf's spec fixes and patched it up so that it builds and runs on rawhide x86_64. New SRPM: http://www.auroralinux.org/people/spot/review/vamos-0.5.7-2.fc8.src.rpm New SPEC: http://www.auroralinux.org/people/spot/review/vamos.spec
This builds for me on rawhide-x86_64, but I get a huge load of rpmlint complaints: This one should be easy to fix: W: vamos-devel wrong-file-end-of-line-encoding /usr/share/doc/vamos-devel-0.5.7/matrixfaq.htm The rest are all undefined-non-weak-symbol warnings, 612 of them. I'll attach them. The mangled symbols make it somewhat difficult to know what's going on, but there are unmangled symbols as well like "tan", "asin", "glGenLists" and "glClear" which make it seem like these libraries aren't even being linked against libm or the GL libraries, much less each other.
Created attachment 158683 [details] undefined-non-weak-symbol warnings
Revisiting this after a long time... the good news is that all the undefined-non-weak-symbol warnings are gone. The bad news is that the code is still a twisted nightmare of GL pain which simply segfaults. I'm no longer interested in trying to get this working, so I'm closing this out.