Arch Linux User Repository

@Beinje unfortunately not (other than what I wrote about below). I was hoping to perhaps hear something from eclairevoyant. I've had to move on from this for the time being. Not sure if it's something that should be taken up with upstream instead...

@berilac I got the same error related to QPBO (undefined references to QPBO<int>::MergeParallelEdges(QPBO<int>::Arc*, QPBO<int>::Arc*) and QPBO<int>::maxflow(bool, bool)) when trying to build the package. Did you manage to solve the issue ?

Tried building outside of a chroot using the same instructions found in the PKGBUILD's build(). Seems to build fine, but oddly considerably faster, by what feels like an order of magnitude. The chroot'ed output is also much more verbose - prints a lot more warnings. Trying to determine what's different.

So far, I've found a difference in the FLAG variable set by cmake.

In the chroot, it is this:

-march=x86-64 -mtune=generic -O2 -pipe -fno-plt -fexceptions -Wp,-D_FORTIFY_SOURCE=3 -Wformat -Werror=format-security -fstack-clash-protection -fcf-protection -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -Wp,-D_GLIBCXX_ASSERTIONS -flto=auto -std=c++11

Outside the chroot it's only this:

-std=c++11

Presumably the compilation speed is at least in part to do with a lack of -O2.

Ok, so it looks like it's building from code in the repo. src/uti_phgrm/GraphCut/QPBO-v1.4. I've had no luck so far determining the source of the problem.

Getting a bunch of linker errors; all (so far) related to QPBO, e.g.:

[ 89%] Linking CXX executable mm3d
...
/usr/bin/ld: /tmp/cc3M52dF.ltrans66.ltrans.o: in function `QPBO<int>::MergeParallelEdges()':
<artificial>:(.text+0x6a81): undefined reference to `QPBO<int>::MergeParallelEdges(QPBO<int>::Arc*, QPBO<int>::Arc*)'

"Full" output (from the same starting point as above) here: https://termbin.com/21y1

I've tried looking it up, with not much clear success. Might it be a missing dependency or could something else be the matter? I'll keep digging myself as well.

@dreieck chill out, no need for the long essay. I already said the same a while back, if you read the comments, just didn't get round to updating this yet.

Anyway it's updated now.

Please remove dependency on python-argparse-shim. This is not needed at all. python-argparse is part of core python. python-argparse-shim is only there for packages which have not been fixed yet and still depend on python-argparse, to fulfill their dependency.

Depending on python-argparse-shim is pointless, superfluos, and might break when python-argparse-shim is to be deleted because it came obsolete.

Thanks for maintaining!

orphaned

@MarsSeed, no, it shouldn't depend on anything related to argparse, because argparse is part of the python language already. Depending on the shim is pointless, the "shim" does absolutely nothing

Please kindly make this depend on python-argparse-shim to eliminate using the deprecated and problematic python-argparse, and letting this application use the natively built-in Python argparse functionality.