Bug#808634: cadabra in Debian
Hi Kasper, Kasper Peeters wrote: > > Uploaded last night. The build failed on a bunch of architectures > > (arm64, armel, armhf, pppc64el, powerpc, s390x)) due to failing tests. > > It was always the test fieldtheory which hang on these architectures > > and was killed after reaching an timeout. > > > > My current suspicion is that lie doesn't behave as expected on these > > architectures. Haven't checked it explicitly, though. > > I would not waste more time on this; the bit that requires LiE is only > a very small part of the total functionality, and this dependency will > go away in 2.x anyway. Just disabling that test is fine with me. That's what I have done and what already helped: All architectures which failed with that test have now passed the test suite and built fine. Additionally all release (i.e. relevant) architectures already built fine, too: https://buildd.debian.org/status/package.php?p=cadabra&suite=unstable So it looks quite good now. Will nevertheless keep an eye on it. Regards, Axel -- ,''`. | Axel Beckert , http://people.debian.org/~abe/ : :' : | Debian Developer, ftp.ch.debian.org Admin `. `' | 4096R: 2517 B724 C5F6 CA99 5329 6E61 2FF9 CD59 6126 16B5 `-| 1024D: F067 EA27 26B9 C3FC 1486 202E C09E 1D89 9593 0EDE
Bug#808634: cadabra in Debian
> Uploaded last night. The build failed on a bunch of architectures > (arm64, armel, armhf, pppc64el, powerpc, s390x)) due to failing tests. > It was always the test fieldtheory which hang on these architectures > and was killed after reaching an timeout. > > My current suspicion is that lie doesn't behave as expected on these > architectures. Haven't checked it explicitly, though. I would not waste more time on this; the bit that requires LiE is only a very small part of the total functionality, and this dependency will go away in 2.x anyway. Just disabling that test is fine with me. Cheers, Kasper
Bug#808634: cadabra in Debian
Hi Kasper, Axel Beckert wrote: > Kasper Peeters wrote: > > > It builds then, but the test suite seems to hang at > > > "fieldtheory" (or I have to wait for more than a few minutes there): > > > > Weird, this test went through fine on a VM which I installed from > > scratch with Debian 8.6 and then upgraded to testing. > > Found the issue: The package didn't declare a build-dependency on lie. > If lie is installed, the test works fine. Uploaded last night. The build failed on a bunch of architectures (arm64, armel, armhf, pppc64el, powerpc, s390x)) due to failing tests. It was always the test fieldtheory which hang on these architectures and was killed after reaching an timeout. My current suspicion is that lie doesn't behave as expected on these architectures. Haven't checked it explicitly, though. So for now, I'll again disable that test to hopefully get it passing the test suite on all architectures. Regards, Axel -- ,''`. | Axel Beckert , http://people.debian.org/~abe/ : :' : | Debian Developer, ftp.ch.debian.org Admin `. `' | 4096R: 2517 B724 C5F6 CA99 5329 6E61 2FF9 CD59 6126 16B5 `-| 1024D: F067 EA27 26B9 C3FC 1486 202E C09E 1D89 9593 0EDE
Bug#808634: cadabra in Debian
Hi Kaspar, Kasper Peeters wrote: > > It builds then, but the test suite seems to hang at > > "fieldtheory" (or I have to wait for more than a few minutes there): > > Weird, this test went through fine on a VM which I installed from > scratch with Debian 8.6 and then upgraded to testing. Found the issue: The package didn't declare a build-dependency on lie. If lie is installed, the test works fine. I think I have most of the packaging updates covered. Will upload soon-ish. Regards, Axel -- ,''`. | Axel Beckert , http://people.debian.org/~abe/ : :' : | Debian Developer, ftp.ch.debian.org Admin `. `' | 4096R: 2517 B724 C5F6 CA99 5329 6E61 2FF9 CD59 6126 16B5 `-| 1024D: F067 EA27 26B9 C3FC 1486 202E C09E 1D89 9593 0EDE
Bug#808634: cadabra in Debian
> It builds then, but the test suite seems to hang at > "fieldtheory" (or I have to wait for more than a few minutes there): Weird, this test went through fine on a VM which I installed from scratch with Debian 8.6 and then upgraded to testing. As this test depends on the 'lie' program, it may well a be a problem in the interaction with that package. For the time being, I am fine with disabling this specific test for Debian testing. I have tagged the github repo with 1.46; have left the test active there. Cheers, Kasper
Processed: Re: Bug#808634: cadabra in Debian
Processing control commands: > tag -1 + pending Bug #808634 [src:cadabra] cadabra: FTBFS: manipulator.hh: fatal error: sigc++/object.h: No such file or directory Added tag(s) pending. -- 808634: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=808634 Debian Bug Tracking System Contact ow...@bugs.debian.org with problems
Bug#808634: cadabra in Debian
Control: tag -1 + pending
Hi Kasper,
[Removing Iulian from Cc. I assume he'll be rather annoyed by this
debugging session.]
Axel Beckert wrote:
> Kasper Peeters wrote:
> > I have finally had the time to install Debian Testing and fix the
> > remaining issues. Version 1.46 is on github.com/kpeeters/cadabra now,
> > and builds/tests without issues with gcc-6.
>
> Much appreciated. Mind pushing the according tag to GitHub, too?
>
> I already see the new release in this commit:
> https://github.com/kpeeters/cadabra/commit/11d95182bf20e62add761801dcafdd908962c735
Actually I was surprised that the patch was so trivial looking. But
I probably would have had a hard time to have guessed that kind of fix
from the error messags G++ threw. Thanks again!
Unfortunately we're back to this issue:
Axel Beckert wrote:
> It builds then, but the test suite seems to hang at "fieldtheory" (or
> I have to wait for more than a few minutes there):
>
> [...]
> running test "gamma"...passed.
> running test "symmetry"...passed.
> running test "fieldtheory"...
>
> Process tree as shown by htop:
>
> 5952 root 20 0 8656 1780 1480 S 0.0 0.0 0:00.00 │ │
> └─ make -j1 test VERBOSE=1
> 5956 root 20 0 8428 776 680 S 0.0 0.0 0:00.00 │ │
> └─ /bin/sh -c ( export CDB_PARANOID=1 && export
> CDB_ERRORS_ARE_FATAL=1 \ && export CDB_PRINTSTAR=1 && cd tests && make
> clean && make all);
> 5957 root 20 0 8656 2036 1648 S 0.0 0.0 0:00.00 │ │
>└─ make all
> 6056 root 20 0 8428 752 652 S 0.0 0.0 0:00.00 │ │
> └─ /bin/sh -c /usr/bin/time -o timing.log -a -f "%U %S"
> ../src/cadabra --bare --input fieldtheory.cdb > fieldtheory.res
> 6057 root 20 0 8324 692 600 S 0.0 0.0 0:00.00 │ │
> └─ /usr/bin/time -o timing.log -a -f %U %S
> ../src/cadabra --bare --input fieldtheory.cdb
> 6058 root 20 0 34112 4464 4136 S 0.0 0.0 0:00.00 │ │
> └─ ../src/cadabra --bare --input fieldtheory.cdb
> 6059 root 20 0 27716 3240 2976 S 0.0 0.0 0:00.00 │ │
>└─ ptywrap lie
> 6060 root 20 0 27716 264 0 S 0.0 0.0 0:00.00 │ │
> └─ ptywrap lie
This part looks still the same.
I've waited at least 15 minutes this time.
An strace on the two ptywrap processes doesn't show much, looks like some
waiting loop:
→ strace -p19718 -p19717
strace: Process 19718 attached
strace: Process 19717 attached
[pid 19717] select(16, [0 11 13 15], NULL, [0 11 13], NULL
[pid 19718] read(15, 0x7ffd45d49140, 1) = ? ERESTARTSYS (To be restarted if
SA_RESTART is set)
[pid 19717] <... select resumed> ) = ? ERESTARTNOHAND (To be restarted if
no handler)
[pid 19717] --- SIGWINCH {si_signo=SIGWINCH, si_code=SI_KERNEL} ---
[pid 19718] --- SIGWINCH {si_signo=SIGWINCH, si_code=SI_KERNEL} ---
[pid 19717] select(16, [0 11 13 15], NULL, [0 11 13], NULL
[pid 19718] read(15,
[pid 19717] <... select resumed> ) = ? ERESTARTNOHAND (To be restarted if
no handler)
[pid 19718] <... read resumed> 0x7ffd45d49140, 1) = ? ERESTARTSYS (To be
restarted if SA_RESTART is set)
[pid 19717] --- SIGWINCH {si_signo=SIGWINCH, si_code=SI_KERNEL} ---
[pid 19718] --- SIGWINCH {si_signo=SIGWINCH, si_code=SI_KERNEL} ---
[pid 19717] select(16, [0 11 13 15], NULL, [0 11 13], NULL
[pid 19718] read(15,
[pid 19717] <... select resumed> ) = ? ERESTARTNOHAND (To be restarted if
no handler)
[pid 19718] <... read resumed> 0x7ffd45d49140, 1) = ? ERESTARTSYS (To be
restarted if SA_RESTART is set)
[pid 19717] --- SIGWINCH {si_signo=SIGWINCH, si_code=SI_KERNEL} ---
[pid 19718] --- SIGWINCH {si_signo=SIGWINCH, si_code=SI_KERNEL} ---
[pid 19717] select(16, [0 11 13 15], NULL, [0 11 13], NULL
[pid 19718] read(15,
[pid 19717] <... select resumed> ) = ? ERESTARTNOHAND (To be restarted if
no handler)
[pid 19718] <... read resumed> 0x7ffd45d49140, 1) = ? ERESTARTSYS (To be
restarted if SA_RESTART is set)
[pid 19717] --- SIGWINCH {si_signo=SIGWINCH, si_code=SI_KERNEL} ---
[pid 19718] --- SIGWINCH {si_signo=SIGWINCH, si_code=SI_KERNEL} ---
[pid 19717] select(16, [0 11 13 15], NULL, [0 11 13], NULL
[pid 19718] read(15,
[pid 19717] <... select resumed> ) = ? ERESTARTNOHAND (To be restarted if
no handler)
[pid 19718] <... read resumed> 0x7ffd45d49140, 1) = ? ERESTARTSYS (To be
restarted if SA_RESTART is set)
[pid 19717] --- SIGWINCH {si_signo=SIGWINCH, si_code=SI_KERNEL} ---
[pid 19718] --- SIGWINCH {si_signo=SIGWINCH, si_code=SI_KERNEL} ---
[pid 19717] select(16, [0 11 13 15], NULL, [0 11 13], NULL
[pid 19718] read(15,
[pid 19717] <... select resumed> ) = ? ERESTARTNOHAND (To be restarted if
no handler)
[pid 19718] <... read resumed> 0x7ffd45d49140, 1) = ? ERESTARTSYS (To be
restarted if SA_RESTART is set)
[pid 19717] -
Bug#808634: cadabra in Debian
Hi Kasper, Kasper Peeters wrote: > I have finally had the time to install Debian Testing and fix the > remaining issues. Version 1.46 is on github.com/kpeeters/cadabra now, > and builds/tests without issues with gcc-6. Much appreciated. Mind pushing the according tag to GitHub, too? I already see the new release in this commit: https://github.com/kpeeters/cadabra/commit/11d95182bf20e62add761801dcafdd908962c735 But it's neither yet visible on https://github.com/kpeeters/cadabra/releases nor on https://github.com/kpeeters/cadabra/tags (I can also do it without that tag, but it would ease some things.) > Many thanks to all for helping out with this. Axel, it would be much > appreciated if you can still push this into the next stable. I'll try and I'm quite optimistic thanks to your prompt action! Regards, Axel -- ,''`. | Axel Beckert , http://people.debian.org/~abe/ : :' : | Debian Developer, ftp.ch.debian.org Admin `. `' | 4096R: 2517 B724 C5F6 CA99 5329 6E61 2FF9 CD59 6126 16B5 `-| 1024D: F067 EA27 26B9 C3FC 1486 202E C09E 1D89 9593 0EDE
Bug#808634: cadabra in Debian
I have finally had the time to install Debian Testing and fix the remaining issues. Version 1.46 is on github.com/kpeeters/cadabra now, and builds/tests without issues with gcc-6. Many thanks to all for helping out with this. Axel, it would be much appreciated if you can still push this into the next stable. Cheers, Kasper
Bug#808634: cadabra in Debian
Hi Kasper, Axel Beckert wrote on 3rd of October: > Kasper Peeters wrote: > > > Were you able to build it with an updated system using GCC 6.2.x? > > > > An 'apt-get update' and 'apt-get dist-upgrade' still does not bring me > > 6.2.x. Do I need to do anything else to get it on testing? > > My fault, sorry: I mixed up the switch from gcc-5 to gcc-6 as default > (which happened in testing in the middle of August) with the switch > from gcc-6 6.1.x to gcc-6 6.2.x which did indeed not yet happen in > testing. > > According to https://packages.qa.debian.org/g/gcc-6.html it might > happen in 2 or 3 days. Were you able to reproduce the build failures in the meanwhile? Unfortunately the deadline for getting into the next Debian stable release is approaching quickly -- and I should do a working upload before 24th of December (end of this week) to have a chance to get into the next Debian Stable release. You don't need to do a new release for that -- a working patch against 1.45 or the right compiler flags to get it compiling would suffice. Regards, Axel -- ,''`. | Axel Beckert , http://people.debian.org/~abe/ : :' : | Debian Developer, ftp.ch.debian.org Admin `. `' | 4096R: 2517 B724 C5F6 CA99 5329 6E61 2FF9 CD59 6126 16B5 `-| 1024D: F067 EA27 26B9 C3FC 1486 202E C09E 1D89 9593 0EDE
Bug#808634: cadabra in Debian
Hi Kasper, Kasper Peeters wrote: > > Were you able to build it with an updated system using GCC 6.2.x? > > An 'apt-get update' and 'apt-get dist-upgrade' still does not bring me > 6.2.x. Do I need to do anything else to get it on testing? My fault, sorry: I mixed up the switch from gcc-5 to gcc-6 as default (which happened in testing in the middle of August) with the switch from gcc-6 6.1.x to gcc-6 6.2.x which did indeed not yet happen in testing. According to https://packages.qa.debian.org/g/gcc-6.html it might happen in 2 or 3 days. Regards, Axel -- ,''`. | Axel Beckert , http://people.debian.org/~abe/ : :' : | Debian Developer, ftp.ch.debian.org Admin `. `' | 4096R: 2517 B724 C5F6 CA99 5329 6E61 2FF9 CD59 6126 16B5 `-| 1024D: F067 EA27 26B9 C3FC 1486 202E C09E 1D89 9593 0EDE
Bug#808634: cadabra in Debian
> Were you able to build it with an updated system using GCC 6.2.x? An 'apt-get update' and 'apt-get dist-upgrade' still does not bring me 6.2.x. Do I need to do anything else to get it on testing? Cheers, Kasper
Bug#808634: cadabra in Debian
Hi Kasper, Axel Beckert wrote: > Kasper Peeters wrote: > > I have just tried building cadabra 1.45 on a fresh install of Testing, > > so that's with gcc 6.1.1 (not using the pbuilder, but direct on the > > hardware). No problem. Any idea why it doesn't build on your machine > > but does on mine? > > Thanks for testing. Will check. I just verified that it still fails to builds for me as mentioned on two different Debian Unstable amd64 machines as well as inside an clean Debian Unstable amd64 chroot (using pbuilder) -- all with GCC 6.2.x. I can also confirm that it built fine with GCC 6.1.1. (I've set up a chroot with all packages in the version when the last GCC 6.1.1 package was uploaded to Debian Unstable, i.e. as of 2016-Aug-16 at 04:30:10 UTC.) It builds then, but the test suite seems to hang at "fieldtheory" (or I have to wait for more than a few minutes there): [...] running test "gamma"...passed. running test "symmetry"...passed. running test "fieldtheory"... Process tree as shown by htop: 5952 root 20 0 8656 1780 1480 S 0.0 0.0 0:00.00 │ │ └─ make -j1 test VERBOSE=1 5956 root 20 0 8428 776 680 S 0.0 0.0 0:00.00 │ │ └─ /bin/sh -c ( export CDB_PARANOID=1 && export CDB_ERRORS_ARE_FATAL=1 \ && export CDB_PRINTSTAR=1 && cd tests && make clean && make all); 5957 root 20 0 8656 2036 1648 S 0.0 0.0 0:00.00 │ │ └─ make all 6056 root 20 0 8428 752 652 S 0.0 0.0 0:00.00 │ │ └─ /bin/sh -c /usr/bin/time -o timing.log -a -f "%U %S" ../src/cadabra --bare --input fieldtheory.cdb > fieldtheory.res 6057 root 20 0 8324 692 600 S 0.0 0.0 0:00.00 │ │ └─ /usr/bin/time -o timing.log -a -f %U %S ../src/cadabra --bare --input fieldtheory.cdb 6058 root 20 0 34112 4464 4136 S 0.0 0.0 0:00.00 │ │ └─ ../src/cadabra --bare --input fieldtheory.cdb 6059 root 20 0 27716 3240 2976 S 0.0 0.0 0:00.00 │ │ └─ ptywrap lie 6060 root 20 0 27716 264 0 S 0.0 0.0 0:00.00 │ │ └─ ptywrap lie > I tried on Unstable and Unstable has gcc 6.2.0 for a few weeks now and > it seems to have migrated to testing just today (around noon > probably), so that could be a potential reason. (And currently the > only one which comes to my mind at the moment.) > > You should get it on Testing now, too, if you do an "apt-get update; > apt-get dist-upgrade" on your Testing machine (and you mirror has > already synced). Were you able to build it with an updated system using GCC 6.2.x? Regards, Axel -- ,''`. | Axel Beckert , http://people.debian.org/~abe/ : :' : | Debian Developer, ftp.ch.debian.org Admin `. `' | 4096R: 2517 B724 C5F6 CA99 5329 6E61 2FF9 CD59 6126 16B5 `-| 1024D: F067 EA27 26B9 C3FC 1486 202E C09E 1D89 9593 0EDE
Bug#808634: cadabra in Debian
Hi Kasper, Kasper Peeters wrote: > I have just tried building cadabra 1.45 on a fresh install of Testing, > so that's with gcc 6.1.1 (not using the pbuilder, but direct on the > hardware). No problem. Any idea why it doesn't build on your machine > but does on mine? Thanks for testing. Will check. I tried on Unstable and Unstable has gcc 6.2.0 for a few weeks now and it seems to have migrated to testing just today (around noon probably), so that could be a potential reason. (And currently the only one which comes to my mind at the moment.) You should get it on Testing now, too, if you do an "apt-get update; apt-get dist-upgrade" on your Testing machine (and you mirror has already synced). Regards, Axel -- ,''`. | Axel Beckert , http://people.debian.org/~abe/ : :' : | Debian Developer, ftp.ch.debian.org Admin `. `' | 4096R: 2517 B724 C5F6 CA99 5329 6E61 2FF9 CD59 6126 16B5 `-| 1024D: F067 EA27 26B9 C3FC 1486 202E C09E 1D89 9593 0EDE
Bug#808634: cadabra in Debian
Hi Axel, I have just tried building cadabra 1.45 on a fresh install of Testing, so that's with gcc 6.1.1 (not using the pbuilder, but direct on the hardware). No problem. Any idea why it doesn't build on your machine but does on mine? Cheers, Kasper
Bug#808634: cadabra in Debian
Hi Kasper, [removing Iulian from the Cc to not bother him with this.] Kasper Peeters wrote: > Shouldn't be too difficult to fix, but I'll need a Debian install here > to be able to play. Tried 'testing' in a VirtualBox this weekend but > that just boots to a black screen. What's the best/recommended way to > get gcc6 on a Debian 8.6 system? Using pbuilder (creates on-demand throw-away chroots of other Debian or Ubuntu releases) is probably the easiest way to build packages: (A "#" indicates that these commands need to be run with root permissions. Using "sudo" should suffice. A "chroot#" indicates that this command should be issues inside the chroot.) # apt-get install pbuilder # pbuilder create (By default it creates a Debian Unstable chroot, otherwise use e.g. "env DIST=testing pbuilder create".) >From there on there is a tar ball of a virgin Debian Unstable (or whatever distribution you requested) available for ondemand extraction. The tar ball is minimal, but targetted towards for building packages (i.e. make, gcc, g++, etc. are already installed) and which gets extracted upon every package build or login. You can use it in multiple ways as it's targetted towards Debian package development and building. But I assume you rather need something to do upstream development, so most automatisms are not that much of help for you. So just logging in into the chroot is probably most suitable for your needs: # pbuilder login W: /root/.pbuilderrc does not exist I: Building the build Environment I: extracting base tarball [/var/cache/pbuilder/sid-base.tgz] I: creating local configuration I: copying local configuration I: mounting /proc filesystem I: mounting /run/shm filesystem I: mounting /dev/pts filesystem I: policy-rc.d already exists I: Obtaining the cached apt archive contents I: entering the shell File extracted to: /var/cache/pbuilder/build//28735 W: no hooks of type F found -- ignoring root@:/# The you either copy or mount your development tree into the chroot (exact path varies, but is shown upon login, see example above). If build dependencies haven't changed much since 1.39, you should get them all with this command inside the chroot: chroot# apt-get build-dep cadabra You also may want to have everything you need to work with autotools. The command chroot# apt-get install dh-autoreconf will get all you need. (dh-autoreconf is a packaging helper which itself you won't need, but it pulls in autoconf, automake, autopoint and libtool, hence is also convenient as meta-package. :-) Beware that after exiting that shell, everything inside the chroot will be removed and lost. (The cached tar ball will not be touched unless explicitly requested. But once downloaded packages for installation will be cached. Use "pbuilder clean" as root to clean up the package cache and potential leftovers in the build directories.) Once after a while you want to upgrade the tar ball to the newest packages, especially those with Debian Unstable or Testing. Use # pbuilder update for that. Hope this helps. Regards, Axel -- ,''`. | Axel Beckert , http://people.debian.org/~abe/ : :' : | Debian Developer, ftp.ch.debian.org Admin `. `' | 4096R: 2517 B724 C5F6 CA99 5329 6E61 2FF9 CD59 6126 16B5 `-| 1024D: F067 EA27 26B9 C3FC 1486 202E C09E 1D89 9593 0EDE
Bug#808634: cadabra in Debian
Shouldn't be too difficult to fix, but I'll need a Debian install here to be able to play. Tried 'testing' in a VirtualBox this weekend but that just boots to a black screen. What's the best/recommended way to get gcc6 on a Debian 8.6 system? Cheers, Kasper On Sat, 17 Sep 2016 02:22:44 +0200 Axel Beckert wrote: > Hi Kasper, > > Axel Beckert wrote: > > Kasper Peeters wrote: > > > > Maybe you can tag them and push the tags (or just push the tags > > > > if you already have the tags in your local working copy. > > > > > > Sorry, forgot to push the tags, now done. > > > > Much appreciated, thanks! Will have a look at it now. > > Unfortunately it still fails to build (with GCC 6). Build log > attached. Hope, this helps. > > Regards, Axel
Bug#808634: cadabra in Debian
Hi Kasper, Axel Beckert wrote: > Kasper Peeters wrote: > > > Maybe you can tag them and push the tags (or just push the tags if you > > > already have the tags in your local working copy. > > > > Sorry, forgot to push the tags, now done. > > Much appreciated, thanks! Will have a look at it now. Unfortunately it still fails to build (with GCC 6). Build log attached. Hope, this helps. Regards, Axel -- ,''`. | Axel Beckert , http://people.debian.org/~abe/ : :' : | Debian Developer, ftp.ch.debian.org Admin `. `' | 4096R: 2517 B724 C5F6 CA99 5329 6E61 2FF9 CD59 6126 16B5 `-| 1024D: F067 EA27 26B9 C3FC 1486 202E C09E 1D89 9593 0EDE cadabra_1.45-1_amd64.build.gz Description: Binary data
Bug#808634: cadabra in Debian
Hi Kasper, Kasper Peeters wrote: > > Maybe you can tag them and push the tags (or just push the tags if you > > already have the tags in your local working copy. > > Sorry, forgot to push the tags, now done. Much appreciated, thanks! Will have a look at it now. Regards, Axel -- ,''`. | Axel Beckert , http://people.debian.org/~abe/ : :' : | Debian Developer, ftp.ch.debian.org Admin `. `' | 4096R: 2517 B724 C5F6 CA99 5329 6E61 2FF9 CD59 6126 16B5 `-| 1024D: F067 EA27 26B9 C3FC 1486 202E C09E 1D89 9593 0EDE
Bug#808634: cadabra in Debian
> Maybe you can tag them and push the tags (or just push the tags if you > already have the tags in your local working copy. Sorry, forgot to push the tags, now done. Kasper
Bug#808634: cadabra in Debian
Hi Kasper, Kasper Peeters wrote: > I have now fixed all compilation issues with recent libraries and > compilers in 1.45, available from http://github.com/kpeeters/cadabra. > This one builds and works on Ubuntu 16.04 with gcc 5.4.0. > > If any of you could bring this back to life in Debian, that would be > much appreciated! Had a look at the list of cadabra releases, but I can't determine which commit you actually consider to be 1.45: Neither on https://github.com/kpeeters/cadabra/releases nor on https://github.com/kpeeters/cadabra/tags I can find a 1.45 version being declared. I found https://github.com/kpeeters/cadabra/commit/f95fe8dd where the version number is changed from 1.44 to 1.45, but the next commit https://github.com/kpeeters/cadabra/commit/ef0a2975 has the commit message "Working again" so I assume the previous commit does not work. Which of these two commits do you consider to be "1.45"? Maybe you can tag them and push the tags (or just push the tags if you already have the tags in your local working copy. Regards, Axel -- ,''`. | Axel Beckert , http://people.debian.org/~abe/ : :' : | Debian Developer, ftp.ch.debian.org Admin `. `' | 4096R: 2517 B724 C5F6 CA99 5329 6E61 2FF9 CD59 6126 16B5 `-| 1024D: F067 EA27 26B9 C3FC 1486 202E C09E 1D89 9593 0EDE
Bug#808634: cadabra in Debian
I have now fixed all compilation issues with recent libraries and compilers in 1.45, available from http://github.com/kpeeters/cadabra. This one builds and works on Ubuntu 16.04 with gcc 5.4.0. If any of you could bring this back to life in Debian, that would be much appreciated! Cheers, Kasper
Bug#808634: cadabra in Debian
Hi Kasper, Kasper Peeters wrote: > > I'll have a look at your fix. Thanks for the help! Unfortunately it fails in the same way as with my suggested fix. > I haven't had a chance to test this with gcc 5 or 6 yet though; GCC 6 is the default in Debian Unstable and Testing for about a month now. > it may need a bit of further tweaking for those compilers. Seems so, yes. Maybe the hints on https://gcc.gnu.org/gcc-6/porting_to.html do help although I found nothing there which relates to the error message I posted in https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=808634#23 Regards, Axel -- ,''`. | Axel Beckert , http://people.debian.org/~abe/ : :' : | Debian Developer, ftp.ch.debian.org Admin `. `' | 4096R: 2517 B724 C5F6 CA99 5329 6E61 2FF9 CD59 6126 16B5 `-| 1024D: F067 EA27 26B9 C3FC 1486 202E C09E 1D89 9593 0EDE
Bug#808634: cadabra in Debian
> I'll have a look at your fix. Thanks for the help! I haven't had a chance to test this with gcc 5 or 6 yet though; it may need a bit of further tweaking for those compilers. It will not work with clang (will fail somewhere else with issues which I solved for 2.x but do not want to backport for lack of time). Cheers, Kasper
Bug#808634: cadabra in Debian
Hi Kasper,
On Tue, Aug 23, 2016 at 08:12:19PM +0200, Kasper Peeters wrote:
> I think the build issues can be solved, looks like a simple libsigc
> version issue.
I got at least a little bit further with this patch:
--- a/src/Makefile.in
+++ b/src/Makefile.in
@@ -21,7 +21,7 @@
#modules/xperm_no_nests.o
SRCS = `find . -name "*.cc"`
-MCFLAGS = @CFLAGS@ -I. -I@top_srcdir@/src `pkg-config modglue --cflags`
+MCFLAGS = @CFLAGS@ -I. -I@top_srcdir@/src `pkg-config modglue --cflags`
`pkg-config --cflags sigc++-1.2`
TIMESTAMP = -D"RELEASE=\"${RELEASE}\"" -D"DATETIME=\"`date | sed -e 's/ /
/'`\"" -DHOSTNAME=\"`hostname`\"
Now it finds sigc++/object.h and some more header files again.
But the build still fails as follows, just later in the build:
In file included from /usr/include/c++/6/set:60:0,
from ./storage.hh:35,
from ./algorithm.hh:26,
from modules/substitute.hh:24,
from modules/substitute.cc:22:
/usr/include/c++/6/bits/stl_tree.h: In instantiation of ‘std::_Rb_tree<_Key,
_Val, _KeyOfValue, _Compare, _Alloc>::iterator std::_Rb_tree<_Key, _Val,
_KeyOfValue, _Compare, _Alloc>::_M_insert_equal(_Arg&&) [with _Arg =
std::pair >,
tree::pre_order_iterator>; _Key = exptree; _Val = std::pair::pre_order_iterator>; _KeyOfValue =
std::_Select1st::pre_order_iterator> >;
_Compare = tree_exact_less_no_wildcards_mod_prel_obj; _Alloc =
std::allocator::pre_order_iterator> >;
std::_Rb_tree<_Key, _Val, _KeyOfValue, _Compare, _Alloc>::iterator =
std::_Rb_tree_iterator::pre_order_iterator> >]’:
/usr/include/c++/6/bits/stl_multimap.h:529:63: required from
‘std::multimap<_Key, _Tp, _Compare, _Alloc>::iterator std::multimap<_Key, _Tp,
_Compare, _Alloc>::insert(_Pair&&) [with _Pair =
std::pair >,
tree::pre_order_iterator>; = void; _Key =
exptree; _Tp = tree::pre_order_iterator; _Compare =
tree_exact_less_no_wildcards_mod_prel_obj; _Alloc =
std::allocator::pre_order_iterator> >;
std::multimap<_Key, _Tp, _Compare, _Alloc>::iterator =
std::_Rb_tree_iterator::pre_order_iterator> >]’
modules/substitute.cc:880:56: required from here
/usr/include/c++/6/bits/stl_tree.h:1891:27: error: no matching function for
call to ‘std::_Rb_tree::pre_order_iterator>, std::_Select1st::pre_order_iterator> >,
tree_exact_less_no_wildcards_mod_prel_obj, std::allocator::pre_order_iterator> >
>::_M_get_insert_equal_pos(std::pair
>, tree::pre_order_iterator>::first_type&)’
= _M_get_insert_equal_pos(_KeyOfValue()(__v));
~~~^~~~
/usr/include/c++/6/bits/stl_tree.h:1833:5: note: candidate:
std::pair
std::_Rb_tree<_Key, _Val, _KeyOfValue, _Compare,
_Alloc>::_M_get_insert_equal_pos(const key_type&) [with _Key = exptree; _Val =
std::pair::pre_order_iterator>; _KeyOfValue =
std::_Select1st::pre_order_iterator> >;
_Compare = tree_exact_less_no_wildcards_mod_prel_obj; _Alloc =
std::allocator::pre_order_iterator> >;
std::_Rb_tree<_Key, _Val, _KeyOfValue, _Compare, _Alloc>::key_type = exptree]
_Rb_tree<_Key, _Val, _KeyOfValue, _Compare, _Alloc>::
^~~
/usr/include/c++/6/bits/stl_tree.h:1833:5: note: no known conversion for
argument 1 from
‘std::pair >,
tree::pre_order_iterator>::first_type {aka
std::_Rb_tree_const_iterator >}’ to ‘const
key_type& {aka const exptree&}’
Makefile:30: recipe for target 'modules/substitute.o' failed
This looks like a issue caused by the switch to GCC 5 or 6 to me, but
at the moment I have no idea what needs to be changed there.
Kind regards, Axel
--
/~\ Plain Text Ribbon Campaign | Axel Beckert
\ / Say No to HTML in E-Mail and News| a...@deuxchevaux.org (Mail)
X See http://www.nonhtmlmail.org/campaign.html | a...@noone.org (Mail+Jabber)
/ \ I love long mails: http://email.is-not-s.ms/ | http://abe.noone.org/ (Web)
Bug#808634: cadabra in Debian
Hi Kasper, On Tue, Aug 23, 2016 at 09:33:09PM +0200, Kasper Peeters wrote: > > Of course it will be different if the cadabra (1.x) package in Debian > > gets back into shape. > > I have just replied to the issue tracker that this can be solved by > replacing > > #include > > with > > #include > > in src/manipulator.hh. Thanks! Seems as if our mails have crossed each other. :-) > This solves the problem without changing the > dependencies. I have pushed 1.44 to the github repository > at http://github.com/kpeeters/cadabra to keep things in sync. > > If anyone can bring this back to life, that would be much > appreciated. I'll have a look at your fix. Thanks for the help! Kind regards, Axel -- /~\ Plain Text Ribbon Campaign | Axel Beckert \ / Say No to HTML in E-Mail and News| a...@deuxchevaux.org (Mail) X See http://www.nonhtmlmail.org/campaign.html | a...@noone.org (Mail+Jabber) / \ I love long mails: http://email.is-not-s.ms/ | http://abe.noone.org/ (Web)
