Reicast-emulator: Fix the snapcraft build

Created on 12 Sep 2018 · 17Comments · Source: reicast/reicast-emulator

snapcraft build is broken because of c++11 usage

Just found out today, we haven't been pushing snapcraft builds since Aug21.

We need to figure out how to fix this

build log
https://build.snapcraft.io/user/reicast/reicast-emulator/331060

Bug Good First Issue Help Wanted

Source

skmp

Most helpful comment

If you specify something like "fixes #1375" in the pr description, the issue is automatically closed when thr pr is merged, tjere is no need for the extra step to close the pr manually.

baka0815 on 21 Sep 2018

👍2

All 17 comments

~I can't see the issue:~

Your session has expired. Reload the page and sign in again.

Forget that...

baka0815 on 12 Sep 2018

I guess it is private to members of the project? Did you try logging in to snapcraft via github?

Anyways

The error seems to be

/build/reicast/parts/reicast/build/shell/linux/../../core/imgread/common.cpp: In function ‘Disc* OpenDisc(const wchar*)’:
/build/reicast/parts/reicast/build/shell/linux/../../core/imgread/common.cpp:134:13: error: ‘nullptr’ was not declared in this scope

I've attached the log file here snapcraft-log.txt

skmp on 12 Sep 2018

Looks like it was this commit: https://github.com/reicast/reicast-emulator/commit/1a4f597bd814915f1bd6f188c575cdfead4c73bc#diff-68060335a9647442c21edb5aa3cefc17

Shouldn't be hard to fix, I guess.

baka0815 on 12 Sep 2018

Impossible. That commit is flawless 🙄

Seriously, though, nullptr should exist on that platform.

AbandonedCart on 13 Sep 2018

It looks like it's not...

baka0815 on 13 Sep 2018

The error says it is undefined, not that it doesn't exist.

AbandonedCart on 13 Sep 2018

However, can't we just skip the initialization and revert to Disc* rv;?

baka0815 on 13 Sep 2018

That could probably work, since a value is always set (preventing it from slipping by uninitialized)

AbandonedCart on 14 Sep 2018

i386 and amd64 builds restored in a functioning status:
- Tried adding the g++ directives "-std=c++0x" and "-std=gnu++0x" to our makefile; all nullptr build issues were solved but I was getting new ones :(
- Tried changing the nullptr value to an initialized nullptr_t or sth; that's the stupidest thing I did. It got me straight up more build issues
- At last, @baka0815 proposal made the build shut up and succeed :D
Snapcraft builds
Branch with the changes
arm64 and armhf build status:
- Still not fixed
I'm writing an article on reicast and snapcraft where a lot of disrespect will take place about the goddamn platform and their "Canonical Distribution of Kubernetes" that takes 28 minutes to build reicast for amd64 targets :$

1138 Schedule of deliverables will be priority next !

Currently investigating the stability of the existing build.

PS: Submitting PR this evening and closing issue

luserx0 on 21 Sep 2018

PR Submitted, issue closed

luserx0 on 21 Sep 2018

🎉1

If you specify something like "fixes #1375" in the pr description, the issue is automatically closed when thr pr is merged, tjere is no need for the extra step to close the pr manually.

baka0815 on 21 Sep 2018

👍2

@baka0815 I was wondering how that happened before. Thanks for that.

AbandonedCart on 21 Sep 2018

I'm getting this same error on an Odroid XU4

g++   -I/home/pigaming/RetroPie-Setup/tmp/build/reicast/shell/linux/../../core -I/home/pigaming/RetroPie-Setup/tmp/build/reicast/shell/linux/../../core/deps -I/home/pigaming/RetroPie-Setup/tmp/build/reicast/shell/linux/../../core/khronos  -D TARGET_BEAGLE -D TARGET_LINUX_ARMELv7 -DARM_HARDFP -fsingle-precision-constant -g -O3 -D RELEASE -c -D USES_HOMEDIR  -frename-registers -fno-strict-aliasing -ffast-math -ftree-vectorize -fomit-frame-pointer  -marm -mfpu=neon -mfloat-abi=hard -funroll-loops -march=armv7ve -mtune=cortex-a15.cortex-a7  -fno-rtti -fpermissive -fno-operator-names -D_GLIBCXX_USE_CXX11_ABI=0 -D USE_EVDEV `pkg-config --cflags alsa` -D USE_ALSA -D USE_OSS -DGLES /home/pigaming/RetroPie-Setup/tmp/build/reicast/shell/linux/../../core/imgread/common.cpp -o obj-dreamcast-odroidxu3/imgread/common.build_obj
/home/pigaming/RetroPie-Setup/tmp/build/reicast/shell/linux/../../core/imgread/common.cpp: In function ‘Disc* OpenDisc(const wchar*)’:
/home/pigaming/RetroPie-Setup/tmp/build/reicast/shell/linux/../../core/imgread/common.cpp:134:13: error: ‘nullptr’ was not declared in this scope
  Disc* rv = nullptr;
             ^
Makefile:384: recipe for target 'obj-dreamcast-odroidxu3/imgread/common.build_obj' failed
make: *** [obj-dreamcast-odroidxu3/imgread/common.build_obj] Error 1
make: *** Waiting for unfinished jobs....
/tmp/ccdTssPj.s: Assembler messages:
/tmp/ccdTssPj.s:173506: Warning: ignoring changed section attributes for .text
/opt/retropie/emulators
Could not successfully build reicast - Dreamcast emulator Reicast (/home/pigaming/RetroPie-Setup/tmp/build/reicast/shell/linux/reicast.elf not found).

Make file is modified to

CXXFLAGS += -fno-rtti -fpermissive -fno-operator-names -D_GLIBCXX_USE_CXX11_ABI=0

blisstik on 14 Oct 2018

The nullptr is still there... I will create a PR for that this evening.

baka0815 on 15 Oct 2018

@baka0815 This issue has been foxed by the PR that fixes snapcraft issues. This is a closed issue.
If you'd like you can review the PR.

luserx0 on 15 Oct 2018

Yes it's in some PR, and no it's not fixed, because the PR is not merged yet!

Otherwise people wouldn't have the problems they are currently having.

baka0815 on 16 Oct 2018

nullptr is removed from master. C++11(+) may be used in the future once the minimum platform is OK'd for Android. Build flags will be set appropriately.