View Issue Details [ Jump to Notes ] | [ Issue History ] [ Print ] | ||||||||
ID | Project | Category | View Status | Date Submitted | Last Update | ||||
---|---|---|---|---|---|---|---|---|---|
0001788 | Spring engine | Linux | public | 2010-01-08 23:37 | 2010-02-07 11:03 | ||||
Reporter | mammadori | ||||||||
Assigned To | Auswaschbar | ||||||||
Priority | normal | Severity | minor | Reproducibility | always | ||||
Status | resolved | Resolution | fixed | ||||||
Product Version | 0.80.5 | ||||||||
Target Version | Fixed in Version | ||||||||
Summary | 0001788: Spring FTBFS on Architectures different than i386 and amd64 | ||||||||
Description | Forwarding the text from bug 563686 on Debian BTS: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=563686 ---- * ---- spring fails to build on architectures other than i386, amd64 because it passes arch-specific flags to gcc: example on alpha: cc1plus: error: unrecognized command line option "-msse" cc1plus: error: unrecognized command line option "-mfpmath=sse" cc1plus: error: unrecognized command line option "-mieee-fp" On armel: Assembler messages: Error: unknown architecture `i686' Error: unrecognized option -march=i686 cc1plus: error: unrecognized command line option "-msse" cc1plus: error: unrecognized command line option "-mfpmath=sse" cc1plus: error: unrecognized command line option "-mieee-fp" See https://buildd.debian.org/~luk/status/package.php?p=spring for other architectures (the problem is the same). ---- * ---- | ||||||||
Additional Information | This bug was first filed on debian BTS, but since is related to the general build system, I'll report there too. | ||||||||
Tags | No tags attached. | ||||||||
Checked infolog.txt for Errors | |||||||||
Attached Files |
|
![]() |
|
Auswaschbar (reporter) 2010-01-09 01:49 |
Fixed, but it should made clear that these builds are not for online play. Also, the march flag you need to fix yourself (-DMARCH_FLAG=xxx) |
mammadori (reporter) 2010-01-22 12:16 |
This do not fixes FTBFS it just help as you noticed, distribution apart -march=i686 should be enabled not when sizeof(int) is 4, but when "grep -qs '^flags.*cmov' /proc/cpuinfo" is true or "uname -m | grep i686" returns true. This obviously is not nice to cross-compilers where the host machine (the target) is different from the build machine. So I suggest anyway to permit an ovveride via a CMake variable which if set either: 1. just skip the test above and do not set a "-march=" 2. permits to specify a "-march=" (so one can specify a prober subarch) |
Auswaschbar (reporter) 2010-01-22 12:28 |
What part of "Also, the march flag you need to fix yourself (-DMARCH_FLAG=xxx)" didn't you understand? |
mammadori (reporter) 2010-01-23 10:16 |
?½What part of "Also, the march flag you need to fix yourself (-DMARCH_FLAG=xxx)" didn't you understand??? I take the above as you were just joking... ;-) I understood what you said and bades on that I just suggested how to fix it properly by setting the -march without looking the sizeof (which is the uncorrect way to set -march). How something in my reply suggested that I didn't understood the reply? If you want to tag this bug as "wontfix", no problem at all, I'm fine. What I do not understand is how a bug report can be fixed without releasign a version of some code which fix the whole bug and not just a 50% of it; I do not undestand also why if one suggest how to fix the remaining 50% if accused of being deaf. Mine was a suggestion on how to preperly fix it instead of saying "it fixed but you need to fix it yourself". Anyway, feel free to close it as you wish, I'll add a patch to fix FTBFS on my packaging even if your tarball does not have one, I reported the issue there not to annoy but to help other tracking the issue and thinking than also other users than debian one could benefit from this bug being fixed. If reporting bugs is annoying I will not do them, I like to play nice with people as soon I understand what please them. Have a nice day. |
Auswaschbar (reporter) 2010-01-23 12:57 |
What is wrong with setting a custom march flag on commandline during cmake configure? Since that possibility has been there for a looong time. |
![]() |
|||
Date Modified | Username | Field | Change |
---|---|---|---|
2010-01-08 23:37 | mammadori | New Issue | |
2010-01-09 01:49 | Auswaschbar | Note Added: 0004476 | |
2010-01-09 01:50 | Auswaschbar | Status | new => resolved |
2010-01-09 01:50 | Auswaschbar | Resolution | open => fixed |
2010-01-09 01:50 | Auswaschbar | Assigned To | => Auswaschbar |
2010-01-22 12:16 | mammadori | Note Added: 0004521 | |
2010-01-22 12:16 | mammadori | Status | resolved => assigned |
2010-01-22 12:16 | mammadori | Resolution | fixed => reopened |
2010-01-22 12:28 | Auswaschbar | Note Added: 0004522 | |
2010-01-22 12:28 | Auswaschbar | Status | assigned => resolved |
2010-01-22 12:28 | Auswaschbar | Resolution | reopened => fixed |
2010-01-23 10:16 | mammadori | Note Added: 0004524 | |
2010-01-23 10:16 | mammadori | Status | resolved => assigned |
2010-01-23 10:16 | mammadori | Resolution | fixed => reopened |
2010-01-23 12:57 | Auswaschbar | Note Added: 0004527 | |
2010-02-07 11:03 | Auswaschbar | Status | assigned => resolved |
2010-02-07 11:03 | Auswaschbar | Resolution | reopened => fixed |