[solved] unable to build dxx-rebirth

Message

Utsuho Reiuji · Post by **Utsuho Reiuji** » Mon Sep 17, 2018 3:40 pm

Hi,
I'm having the following problem ever since the descent rebirth package was changed:

>>> Emerging (1 of 1) games-action/dxx-rebirth-0.59.100::gentoo
>>> Failed to emerge games-action/dxx-rebirth-0.59.100, Log file:
>>>  '/var/tmp/portage/games-action/dxx-rebirth-0.59.100/temp/build.log'
>>> Jobs: 0 of 1 complete, 1 failed                 Load avg: 1.55, 2.79, 3.62
 * Package:    games-action/dxx-rebirth-0.59.100
 * Repository: gentoo
 * Maintainer: games@gentoo.org totktonada.ru@gmail.com
 * USE:        abi_x86_64 amd64 data descent1 descent2 elibc_glibc ipv6 kernel_linux l10n_de music opengl sc55-musicpack textures userland_GNU
 * FEATURES:   preserve-libs sandbox userpriv usersandbox
>>> Unpacking source...
>>> Unpacking dxx-rebirth_v0.59.100-src.tar.gz to /var/tmp/portage/games-action/dxx-rebirth-0.59.100/work
>>> Unpacking d1xr-sc55-music.dxa to /var/tmp/portage/games-action/dxx-rebirth-0.59.100/work
unpack d1xr-sc55-music.dxa: file format not recognized. Ignoring.
>>> Unpacking d2xr-sc55-music.dxa to /var/tmp/portage/games-action/dxx-rebirth-0.59.100/work
unpack d2xr-sc55-music.dxa: file format not recognized. Ignoring.
>>> Unpacking d1xr-briefings-ger.dxa to /var/tmp/portage/games-action/dxx-rebirth-0.59.100/work
unpack d1xr-briefings-ger.dxa: file format not recognized. Ignoring.
>>> Unpacking d2xr-briefings-ger.dxa to /var/tmp/portage/games-action/dxx-rebirth-0.59.100/work
unpack d2xr-briefings-ger.dxa: file format not recognized. Ignoring.
>>> Unpacking d1xr-hires.dxa to /var/tmp/portage/games-action/dxx-rebirth-0.59.100/work
unpack d1xr-hires.dxa: file format not recognized. Ignoring.
>>> Source unpacked in /var/tmp/portage/games-action/dxx-rebirth-0.59.100/work
>>> Preparing source in /var/tmp/portage/games-action/dxx-rebirth-0.59.100/work/dxx-rebirth_v0.59.100-src ...
 * Applying dxx-rebirth-0.59.100-flags.patch ...
 [ ok ]
 * Applying dxx-rebirth-0.59.100-sharepath.patch ...
 [ ok ]
>>> Source prepared.
>>> Configuring source in /var/tmp/portage/games-action/dxx-rebirth-0.59.100/work/dxx-rebirth_v0.59.100-src ...
>>> Source configured.
>>> Compiling source in /var/tmp/portage/games-action/dxx-rebirth-0.59.100/work/dxx-rebirth_v0.59.100-src ...
scons -j5 prefix=/usr d1x=1 d2x=1 debug=0 ipv6=1 opengl=1 sdlmixer=1 verbosebuild=1
scons: Reading SConscript files ...
===== D1X-Rebirth v0.59.100 =====
===== D2X-Rebirth v0.59.100 =====
DXX-Archive.3: compiling on linux2 for linux into .
DXX-Archive.3: checking version of C++ compiler 'x86_64-pc-linux-gnu-g++' ... 'x86_64-pc-linux-gnu-g++ (Gentoo 7.3.0-r3 p1.4) 7.3.0'
DXX-Archive.3: checking path to assembler ... '/usr/lib/gcc/x86_64-pc-linux-gnu/7.3.0/../../../../x86_64-pc-linux-gnu/bin/as'
DXX-Archive.3: checking version of assembler '/usr/lib/gcc/x86_64-pc-linux-gnu/7.3.0/../../../../x86_64-pc-linux-gnu/bin/as' ... 'GNU assembler (Gentoo 2.31.1 p1) 2.31.1'
DXX-Archive.3: checking path to linker ... '/usr/lib/gcc/x86_64-pc-linux-gnu/7.3.0/../../../../x86_64-pc-linux-gnu/bin/ld'
DXX-Archive.3: checking version of linker '/usr/lib/gcc/x86_64-pc-linux-gnu/7.3.0/../../../../x86_64-pc-linux-gnu/bin/ld' ... 'GNU ld (Gentoo 2.31.1 p1) 2.31.1'
DXX-Archive.3:  CXX: 'x86_64-pc-linux-gnu-g++'
DXX-Archive.3:  CPPDEFINES: None
DXX-Archive.3:  CPPPATH: ['common/include', 'common/main', '.', '']
DXX-Archive.3:  CPPFLAGS: ['-Wno-sign-compare']
DXX-Archive.3:  CXXFLAGS: ['-ftabstop=4', '-Wall', '-Wshadow', '-Werror=extra', '-Werror=format=2', '-Werror=missing-braces', '-Werror=missing-include-dirs', '-Werror=uninitialized', '-Werror=undef', '-Werror=pointer-arith', '-Werror=cast-qual', '-Werror=missing-declarations', '-Werror=redundant-decls', '-Werror=vla', '-march=amdfam10', '-O2', '-pipe', '-pthread', '-funsigned-char', '-march=amdfam10', '-O2', '-pipe']
DXX-Archive.3:  LIBS: None
DXX-Archive.3:  LINKFLAGS: ['-Wl,-O1', '-Wl,--as-needed', '-Wl,-O1', '-Wl,--as-needed']
DXX-Archive.3:  $CCACHE_PREFIX: None
DXX-Archive.3:  $DISTCC_HOSTS: None
DXX-Archive.3: checking whether C++ compiler and linker work...yes
DXX-Archive.3: checking whether C++ compiler accepts -std=gnu++14...yes
DXX-Archive.3: checking endian to use...little
DXX-Archive.3: checking whether to enable word alignment fixups...no
DXX-Archive.3: building with OpenGL
DXX-Archive.3: checking whether to enable release options...yes
DXX-Archive.3: checking whether to enable memory allocation tracking...no
DXX-Archive.3: checking whether to enable level editor...no
DXX-Archive.3: checking whether to enable IPv6 support...yes
DXX-Archive.3: checking whether to enable multiplayer over UDP...yes
DXX-Archive.3: checking whether to enable UDP game tracker...yes
DXX-Archive.3: checking for usable library physfs...no
DXX-Archive.3: checking for usable header physfs.h...no
DXX-Archive.3: checking for parseable header physfs.h...yes
scons: *** Header physfs.h is parseable, but cannot compile the test program.  See sconf.log for details.  Stop.
 * ERROR: games-action/dxx-rebirth-0.59.100::gentoo failed (compile phase):
 *   escons failed.
 * 
 * Call stack:
 *     ebuild.sh, line  124:  Called src_compile
 *   environment, line 2128:  Called escons 'prefix=/usr' 'd1x=1' 'd2x=1' 'debug=0' 'ipv6=1' 'opengl=1' 'sdlmixer=1' 'verbosebuild=1'
 *   environment, line  938:  Called die
 * The specific snippet of code:
 *                   die -n "escons failed."
 * 
 * If you need support, post the output of `emerge --info '=games-action/dxx-rebirth-0.59.100::gentoo'`,
 * the complete build log and the output of `emerge -pqv '=games-action/dxx-rebirth-0.59.100::gentoo'`.
 * The complete build log is located at '/var/tmp/portage/games-action/dxx-rebirth-0.59.100/temp/build.log'.
 * The ebuild environment file is located at '/var/tmp/portage/games-action/dxx-rebirth-0.59.100/temp/environment'.
 * Working directory: '/var/tmp/portage/games-action/dxx-rebirth-0.59.100/work/dxx-rebirth_v0.59.100-src'
 * S: '/var/tmp/portage/games-action/dxx-rebirth-0.59.100/work/dxx-rebirth_v0.59.100-src'

 * Messages for package games-action/dxx-rebirth-0.59.100:

 * ERROR: games-action/dxx-rebirth-0.59.100::gentoo failed (compile phase):
 *   escons failed.
 * 
 * Call stack:
 *     ebuild.sh, line  124:  Called src_compile
 *   environment, line 2128:  Called escons 'prefix=/usr' 'd1x=1' 'd2x=1' 'debug=0' 'ipv6=1' 'opengl=1' 'sdlmixer=1' 'verbosebuild=1'
 *   environment, line  938:  Called die
 * The specific snippet of code:
 *                   die -n "escons failed."

I'm not sure what to look for in the sconf.log. The useflags were

Code: Select all

USE="data descent1 descent2 ipv6 music opengl sc55-musicpack textures -debug -opl3-musicpack"

LuxJux · Post by **LuxJux** » Mon Sep 17, 2018 7:27 pm

You will also need the Original-Game

https://www.dxx-rebirth.com/game-content/

Utsuho Reiuji · Post by **Utsuho Reiuji** » Mon Sep 17, 2018 8:27 pm

I do have the cd roms, but no more rom drive (I ripped both before my drive was not accessible anymore). But now that you mention it, I used to have d1x-rebirth and d2x-rebirth before dxx-rebirth. They worked, but an update required me to uninstall both. So where am I supposed to place the game files? The previous ebuild required me to enter a path to those files...

LuxJux · Post by **LuxJux** » Mon Sep 17, 2018 10:04 pm

https://wiki.ubuntuusers.de/Spiele/Descent/ and https://www.holarse-linuxgaming.de/wiki/descent should help a little bit, ( german guides )

Post by Hu » Tue Sep 18, 2018 1:00 am

From reading the ebuild, it looks like game data is now handled through

Code: Select all

|| ( games-action/descent1-data games-action/descent1-demodata )
|| ( games-action/descent2-data games-action/descent2-demodata )

The immediate problem is games-action/dxx-rebirth-0.59.100 fails to compile with dev-games/physfs-3.0.1-r1, which was reported 2018-04-12 and never fixed. As best I can tell, it was never forwarded upstream (or if it was, I cannot find the report upstream, nor does the Gentoo bug report reference sending it upstream). However, an Arch Linux user reported this upstream as Scons fails with error "Header physfs.h is parseable, but cannot compile the test program." on 2017-10-21 (yes, months before it was reported to Gentoo) and upstream responded the same day with follow up questions, and the next day with Suppress PhysFS deprecation errors, which the Arch user who reported the problem says fixed it for him.

So your problem is that you need Gentoo to bump the ebuild to a newer release, or to a live snapshot if upstream has not provided a fixed release. Interestingly, upstream seems to maintain their own ebuilds, separate from the ones in Gentoo. If the Gentoo ebuild doesn't work for you, you could try copying the upstream ebuild to an overlay. If that works better, ask the Gentoo maintainer to sync to upstream's ebuild. If they're already doing the work, Gentoo may as well benefit from it.

Utsuho Reiuji · Post by **Utsuho Reiuji** » Tue Sep 18, 2018 5:36 pm

hmm, I see. Depending on how much time I will have the next weeks, I'll try to get this to run. With the links you provided, I might be able to troubleshoot the error.

Post by Hu » Wed Sep 19, 2018 12:45 am

What is there to troubleshoot? Upstream was notified and issued a workaround/fix. All that you need to do is use it, either by backporting their fix or updating to a version that includes it. If the fix works for you and you have a Gentoo Bugzilla account, you could post to that bug to remind the Gentoo maintainers that an update is needed.

Utsuho Reiuji · Post by **Utsuho Reiuji** » Sun Sep 23, 2018 1:11 am

I tried the following:

downgrading physfs to v2.* as mentioned in the bugzilla
patching physfs-3.* as mentioned on github

Unfortunately, both did not work out, downgrading will cause dxx-rebirth to exit compiling with the same error and the patch breaks physfs. I'll try some more things once I'm not so tired anymore [/list]

Post by Hu » Sun Sep 23, 2018 4:04 pm

What did you patch in physfs? The github issue only talks about patching dxx-rebirth.

Since Gentoo generally follows an upstream-first philosophy, I suggest you try upstream's code as-is from a more recent ebuild. If it works, bumping the Gentoo ebuild would fix this for everyone. If it doesn't work, upstream needs to be notified so that they can try again.

Utsuho Reiuji · Post by **Utsuho Reiuji** » Sun Sep 23, 2018 5:17 pm

Hu wrote:What did you patch in physfs? The github issue only talks about patching dxx-rebirth.

Since Gentoo generally follows an upstream-first philosophy, I suggest you try upstream's code as-is from a more recent ebuild. If it works, bumping the Gentoo ebuild would fix this for everyone. If it doesn't work, upstream needs to be notified so that they can try again.

Oh, I misunderstood the comments in github then. I applied the patch successfully to dxx-rebirth, but it still fails to build.

Post by Hu » Sun Sep 23, 2018 5:54 pm

From the information available, there is nothing further we can do to help you. If you need help, you must post output showing how it fails, and preferably also tell us exactly what you patched and how.

Utsuho Reiuji · Post by **Utsuho Reiuji** » Sun Sep 23, 2018 9:33 pm

The patch I applied from https://github.com/dxx-rebirth/dxx-rebirth/issues/352:

Code: Select all

diff --git a/SConstruct b/SConstruct
index 560a6ef1a..3c049517e 100644
--- a/SConstruct
+++ b/SConstruct
@@ -3811,6 +3811,7 @@ class DXXCommon(LazyObjectConstructor):
 			CXXFLAGS = ['-funsigned-char'],
 			CPPPATH = ['common/include', 'common/main', '.'],
 			CPPFLAGS = SCons.Util.CLVar('-Wno-sign-compare'),
+			CPPDEFINES = [('PHYSFS_DEPRECATED', '')],
 		)
 		add_flags = defaultdict(list)
 		if user_settings.builddir:

Output for emerging dxx-rebirth:

Code: Select all

x86_64-pc-linux-gnu-g++ -c -ftabstop=4 -Wall -Wshadow -Werror=extra -Werror=format=2 -Werror=missing-braces -Werror=missing-include-dirs -Werror=uninitialized -Werror=undef -Werror=pointer-arith -Werror=cast-qual -Werror=missing-declarations -Werror=redundant-decls -Werror=vla -pthread -funsigned-char -march=amdfam10 -O2 -pipe -std=gnu++14 -Werror=unused -Werror=useless-cast -fvisibility=hidden -Wsuggest-attribute=noreturn -Wlogical-op -Wold-style-cast -Wno-sign-compare -DPHYSFS_DEPRECATED= -DNDEBUG -DRELEASE -D_REENTRANT -DDXX_BUILD_DESCENT_II -D__STDC_FORMAT_MACROS -Icommon/include -Icommon/main -I. -I/usr/include/SDL -Id2x-rebirth/main similar/2d/pcx.cpp -o similar/2d/.d2x-rebirth.pcx.o
In file included from common/include/dxxerror.h:30:0,
                 from similar/main/fireball.cpp:32:
In function ‘dcx::vms_vector& dcx::vm_vec_sub(dcx::vms_vector&, const dcx::vms_vector&, const dcx::vms_vector&)’,
    inlined from ‘d2x::objptridx_t d2x::object_create_explosion_sub(d2x::objptridx_t, d2x::vsegptridx_t, const dcx::vms_vector&, dcx::fix, int, dcx::fix, dcx::fix, dcx::fix, d2x::cobjptridx_t)’ at similar/main/fireball.cpp:150:102:
./dxxsconf.h:53:3: error: call to ‘dcx::vm_vec_sub_same_values’ declared with attribute error: vm_vec_sub with equal value inputs
  F(); \
common/include/vecmat.h:275:3: note: in expansion of macro ‘DXX_ALWAYS_ERROR_FUNCTION’
   DXX_ALWAYS_ERROR_FUNCTION(vm_vec_sub_same_values, "vm_vec_sub with equal value inputs");
   ^~~~~~~~~~~~~~~~~~~~~~~~~
In function ‘dcx::vms_vector& dcx::vm_vec_sub(dcx::vms_vector&, const dcx::vms_vector&, const dcx::vms_vector&)’,
    inlined from ‘void d2x::do_exploding_wall_frame()’ at common/include/vecmat.h:284:39:
./dxxsconf.h:53:3: error: call to ‘dcx::vm_vec_sub_same_values’ declared with attribute error: vm_vec_sub with equal value inputs
  F(); \
common/include/vecmat.h:275:3: note: in expansion of macro ‘DXX_ALWAYS_ERROR_FUNCTION’
   DXX_ALWAYS_ERROR_FUNCTION(vm_vec_sub_same_values, "vm_vec_sub with equal value inputs");
   ^~~~~~~~~~~~~~~~~~~~~~~~~
In function ‘dcx::vms_vector& dcx::vm_vec_sub(dcx::vms_vector&, const dcx::vms_vector&, const dcx::vms_vector&)’,
    inlined from ‘void d2x::do_exploding_wall_frame()’ at common/include/vecmat.h:284:39:
./dxxsconf.h:53:3: error: call to ‘dcx::vm_vec_sub_same_values’ declared with attribute error: vm_vec_sub with equal value inputs
  F(); \
common/include/vecmat.h:275:3: note: in expansion of macro ‘DXX_ALWAYS_ERROR_FUNCTION’
   DXX_ALWAYS_ERROR_FUNCTION(vm_vec_sub_same_values, "vm_vec_sub with equal value inputs");
   ^~~~~~~~~~~~~~~~~~~~~~~~~
x86_64-pc-linux-gnu-g++ -c -ftabstop=4 -Wall -Wshadow -Werror=extra -Werror=format=2 -Werror=missing-braces -Werror=missing-include-dirs -Werror=uninitialized -Werror=undef -Werror=pointer-arith -Werror=cast-qual -Werror=missing-declarations -Werror=redundant-decls -Werror=vla -pthread -funsigned-char -march=amdfam10 -O2 -pipe -std=gnu++14 -Werror=unused -Werror=useless-cast -fvisibility=hidden -Wsuggest-attribute=noreturn -Wlogical-op -Wold-style-cast -Wno-sign-compare -DPHYSFS_DEPRECATED= -DNDEBUG -DRELEASE -D_REENTRANT -DDXX_BUILD_DESCENT_II -D__STDC_FORMAT_MACROS -Icommon/include -Icommon/main -I. -I/usr/include/SDL -Id2x-rebirth/main similar/arch/sdl/digi.cpp -o similar/arch/sdl/.d2x-rebirth.digi.o
scons: *** [similar/main/.d2x-rebirth.fireball.o] Error 1
scons: building terminated because of errors.
 * ERROR: games-action/dxx-rebirth-0.59.100::gentoo failed (compile phase):
 *   escons failed.
 * 
 * Call stack:
 *     ebuild.sh, line  124:  Called src_compile
 *   environment, line 2128:  Called escons 'prefix=/usr' 'd1x=1' 'd2x=1' 'debug=0' 'ipv6=1' 'opengl=1' 'sdlmixer=1' 'verbosebuild=1'
 *   environment, line  938:  Called die
 * The specific snippet of code:
 *                   die -n "escons failed."
 * 
 * If you need support, post the output of `emerge --info '=games-action/dxx-rebirth-0.59.100::gentoo'`,
 * the complete build log and the output of `emerge -pqv '=games-action/dxx-rebirth-0.59.100::gentoo'`.
 * The complete build log is located at '/var/tmp/portage/games-action/dxx-rebirth-0.59.100/temp/build.log'.
 * The ebuild environment file is located at '/var/tmp/portage/games-action/dxx-rebirth-0.59.100/temp/environment'.
 * Working directory: '/var/tmp/portage/games-action/dxx-rebirth-0.59.100/work/dxx-rebirth_v0.59.100-src'
 * S: '/var/tmp/portage/games-action/dxx-rebirth-0.59.100/work/dxx-rebirth_v0.59.100-src'

 * Messages for package games-action/dxx-rebirth-0.59.100:

 * User patches applied.
 * ERROR: games-action/dxx-rebirth-0.59.100::gentoo failed (compile phase):
 *   escons failed.
 * 
 * Call stack:
 *     ebuild.sh, line  124:  Called src_compile
 *   environment, line 2128:  Called escons 'prefix=/usr' 'd1x=1' 'd2x=1' 'debug=0' 'ipv6=1' 'opengl=1' 'sdlmixer=1' 'verbosebuild=1'
 *   environment, line  938:  Called die
 * The specific snippet of code:
 *                   die -n "escons failed."

And the corresponding sconf.log:
https://pastebin.com/vYCvcikd

Post by Hu » Sun Sep 23, 2018 10:48 pm

OK, that's a different issue. According to Google: "vm_vec_sub with equal value inputs" this is similar to Rebirth issue #337: Arch Linux gcc-7.1 fails to build: __builtin_constant_p quirks; -Werror=format-truncation and/or Rebirth issue #385: Build Failure: draw.cpp - error: call to ‘dcx::vm_vec_sub_same_values’ declared with attribute error. The exact file is different, but since it's a header, that's plausible. Upstream fixed #337 in mid-2017. Upstream resolved #385 without any changes, but it looks like that is because that user, like you, tried to compile a version that lacked the fix from #337. It looks like 0.59.100 needs multiple fixes for current systems. How many will be required is unknown, but you may be at this a while - or you could do as I suggested above and try a more recent snapshot, since upstream does seem to fix these problems as they are reported. Is there a specific reason you want to use 0.59.100 instead of a more recent snapshot?

Utsuho Reiuji · Post by **Utsuho Reiuji** » Mon Sep 24, 2018 6:41 pm

that is the most current ebuild available via portage.
I think I could compile the most recent source code, but I have no idea how to use scons..

Code: Select all

scons: Reading SConscript files ...
NameError: name 'srcdir' is not defined:
  File "/home/yuki/reactor/dxx-rebirth_v0.60-weekly-04-14-18-src/SConstruct", line 4019:
    class DXXArchive(DXXCommon):
  File "/home/yuki/reactor/dxx-rebirth_v0.60-weekly-04-14-18-src/SConstruct", line 4067:
    'misc/vgwphys.cpp',
  File "/home/yuki/reactor/dxx-rebirth_v0.60-weekly-04-14-18-src/SConstruct", line 4026:
    __get_objects_common=DXXCommon.create_lazy_object_getter([os.path.join(srcdir, f) for f in [

There was a similar or same issue:
https://github.com/dxx-rebirth/dxx-rebirth/issues/350
But the maintainer said they fixed everything so I have no clue what is wrong now..

Post by Hu » Tue Sep 25, 2018 1:12 am

Yes, the ebuild in Portage is out of date for this package. The Gentoo maintainer for this package seems to be on hiatus, as evidenced by the bug report from April that was already solved upstream before it was filed in Gentoo, yet is still open in Gentoo and was never referred upstream. That is why I encouraged you to use upstream's ebuilds, which appear to be much more current. Either those work and solve your problem, or they don't work and somebody needs to report the breakage so that they get fixed. Once they work, whether now or after reporting any lingering problems, someone can ask the Gentoo maintainer to update to a newer version.

As I read the cited ticket, the developer closed it as not-reproducible in October 2017. In June 2018, he said he fixed it and cited some commits that are all dated April 30th 2018. You tried to build a snapshot from April 14th (assuming American MM-DD-YY, rather than European DD-MM-YY, since the American interpretation is the only interpretation that does not cause an out of range month). The April 14th snapshot would predate the cited commits by about two weeks. Looking at the change in the first commit cited versus the line printed in the traceback you showed also indicates that you don't have that fix. Please try applying the developer's fixes or, preferably, using current code.

Utsuho Reiuji · Post by **Utsuho Reiuji** » Tue Sep 25, 2018 8:30 pm

hmm, I didn't realise that the nightly builds were not the most recent. Building the cloned git seems to work.
Thanks for the help, Hu ^^

Post by Hu » Wed Sep 26, 2018 1:00 am

Did you build from source by hand or did you use the ebuild I referenced above? Since the Gentoo maintained ebuild for this package no longer works as-is, it needs to be modified or replaced. Ideally, upstream's ebuilds would be in good enough condition that the Gentoo maintainers could replace the current ebuild with the ones from upstream. If you used the ebuilds I referenced, and they worked, that would be helpful in getting those ebuilds into the tree. If you have not tried them, someone will need to try them before they can be added to Portage.

Regardless, I suggest filing a bug requesting that the Gentoo maintainers bump to a newer version. Users should be able to run emerge dxx-rebirth and get a working install, without local patches or other special workarounds. Currently, that does not happen. You hit 3 different problems (physfs 3, gcc-7, Python 3) with the current snapshot in the tree, and the goal is 0 problems.