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
STK 0.9 fails to build on armel/armhf/arm64/ppc64el/s390x #2194
Comments
|
Here is similar bug report:
#2119
It seems to be something wrong with compiler definitions used in
angelscript because function CallSystemFunctionNative is available in
as_callfunc_arm.cpp and we get undefined reference... But I had only a very
quick look on it.
Btw. I noticed that you are trying to use system glew library. AFAIK we
still need patched glew to make it work properly. @hiker do you remember
what is precisely patched and why do we need it?
|
|
Was about to file a Debian bug when I noticed this one :) All the 64-bit failures are caused by broken architecture detection here: It assumes that x86_64 is the only arch which defines |
|
Could you perhaps test ir #if x86_64 works better? You seem to have access to more platforms than I do |
|
@jcowgill Admittedly I'm not all that familiar with porting to exotic archs, and don't know what preprocessor checks should be used instead; would you care to suggest the correct defines to be used here (or perhaps a patch)? Thanks! @auriamg Sure, I'll prepare a PR once I've run a few test builds on some Debian porterboxen. @deveee Thanks for the heads up re: glew. Interested in seeing what @hiker has to say, since I need some rationale before overriding lintian (glew happens to be one of the libs that lintian will detect when checking for embedded libraries) and choosing to use embedded libs. |
…ixes supertuxkart#2194 For the build failures on 64-bit arches, angelscript was using __LP64__ to test for x86_64. Use the correct define instead. For ARM architectures, ensure the native function calling file is built on all ARM arches and not just when building for android.
|
I submitted #2195, hopefully it fixes this. |
…ixes supertuxkart#2194 For the build failures on 64-bit arches, angelscript was using __LP64__ to test for x86_64. Use the correct define instead. For ARM architectures, ensure the native function calling file is built on all ARM arches and not just when building for android. Also add the nesseary -Wa,-mimplicit-it option when building the assembly file.
|
After verifying, these architectures actually do not appear to be supported by angelscript ( see http://www.angelcode.com/angelscript/features.html ) - they are also pretty uncommon architectures so I don't think we care STK not running there |
|
Re libglew: afaik the problem with libglew still exists. A patch has been supplied (2 years ago). Details in https://sourceforge.net/p/glew/patches/40/ So, till this is fixed upstream, we need our own patched libglew version. |
|
From the angelcode features page:
So it should work on these arches. |
|
Angelscript has been updated to a newer version, should hopefully be fixed now |
STK 0.9 (just uploaded today to Debian) fails to build on a variety of non-i386/amd64 architectures in Debian sid; see here for build status and logs (you can click on any of the rows under the "Status" column for the entire log for that arch). Note that STK 0.8.1 built fine on all archs supported by Debian.
All of the build failures have this in common:
Does angelscript require porting assembly to various non-x86 archs? The weird thing is that STK 0.9 actually builds fine on some non-x86 archs, e.g. mipsel, and I'd be kinda surprised if some piece of software supported mips but not arm these days. However, it does in fact look like angelscript has been ported to arm (lib/angelscript/source/as_callfunc_arm.cpp); why is x86 assembly being used in the arm build then?
It seems to me that this is probably related to the build system somehow, but I'm not all that familiar with cmake; does anyone know what might be going on here?
The text was updated successfully, but these errors were encountered: