fix or remove cliquer

[vbraun]

The installation of cliquer can fail randomly with:

error: /Applications/Xcode.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/bin/install_name_tool: changing install names or rpaths can't be redone for: /Users/vbraun/Code/binary-pkg/source/SageMath/jc4b6yulaujayb9sr94ia88eourzeqip0oidmas391yaj24ng0bmdu/local/lib/libcliquer.dylib (for architecture x86_64) because larger updated load commands do not fit (the program must be relinked, and you may need to use -headerpad or -headerpad_max_install_names)

Really the current cliquer has no business being a standard package:

• No build system to speak of

get autotoolized tarball at http://users.ox.ac.uk/~coml0531/sage/cliquer-1.21.tar.gz

• Errors are not caught
  (probably another ticket?)

• A libcliquer.so is installed on Darwin (in addition to libcliquer.dylib)
  (not with the new autotoolized version)

Upstream: Reported upstream. Developers deny it's a bug.

CC: @dimpase @nathanncohen @williamstein

Component: packages: standard

Author: Dima Pasechnik

Branch/Commit: 8d7f039

Reviewer: Volker Braun

Issue created by migration from https://trac.sagemath.org/ticket/19967

[dimpase]

comment:2

I'll let it burn in autohell, to begin with. Wish me luck. Hopefully this at least makes it installable everywhere we need...

[kiwifb]

comment:3

Replying to @dimpase:

I'll let it burn in autohell, to begin with. Wish me luck. Hopefully this at least makes it installable everywhere we need...

luck.

[dimpase]

comment:4

https://github.com/dimpase/autocliquer has the dirty stuff done; one can build shared libs on Linux and OSX and cliquer's own tests pass. I don't know how to properly hook the tests into the autohell (so that make check does the right thing), please advice (or submit a pull request :-))

make check should run test/testcases, reading data from the same directory.

what the the right Makefile.am magic for them?

Similarly, there are include/cliquer/test* files that should not really be installed, they only needed for make check

I can't also figure out how to set up the proper version number on the .so file. It's again something for configure.ac or one of Makefile.am.

[kiwifb]

comment:5

For proper version number it is in configure.ac and that's libtool stuff. Look at eclib for a nice example https://github.com/JohnCremona/eclib/blob/master/configure.ac I would recommend 1.0.0 to start with.
for the rest I imagine you'll take PR if we send them.

[vbraun]

comment:6

PR sent...

[jdemeyer]

comment:7

Replying to @vbraun:

PR sent...

Which PR to who?

[dimpase]

comment:8

Replying to @jdemeyer:

Replying to @vbraun:

PR sent...

Which PR to who?

a PR to me on the github repo https://github.com/dimpase/autocliquer
(thanks Volker!)

I have already used it and make a proper testsuite; tests pass. Now to the version thing, and it will be ready for making its way into Sage.

[dimpase]

New commits:

b736207	trac #19777: Three new SRGs and db update
5aca3d8	trac #19777: Convert the other 2-intersection set
6696d1e	Merge branch 'develop' of git://trac.sagemath.org/sage into develop
c5374d8	Merge branch 'develop' of git://trac.sagemath.org/sage into develop
b88f7e8	Merge branch 'develop' of git://trac.sagemath.org/sage into develop
136628a	Merge branch 'develop' of git://trac.sagemath.org/sage into develop
05c5881	changes for autotoolozed version

[dimpase]

Branch: public/autocliquer

[dimpase]

Commit: 05c5881

[dimpase]

comment:10

oh, sorry, the same rotten develop branch of me --- will be fixed soon

[dimpase]

comment:11

you may meanwhile just review the last commit (​05c5881)

sorry for mess

[sagetrac-git]

Branch pushed to git repo; I updated commit sha1. This was a forced push. New commits:

3c17486

changes for autotoolozed version

[sagetrac-git]

Changed commit from 05c5881 to 3c17486

[jdemeyer]

comment:13

Why fork cliquer? Did you even try to report this upstream?

[dimpase]

comment:14

Replying to @jdemeyer:

Why fork cliquer? Did you even try to report this upstream?

I know Patric, the maintainer, for over 20 years. I sent him an email, but he is a busy guy, and I am not sure how quickly he will react. Meanwhile this is all we have.

[dimpase]

Upstream: Reported upstream. No feedback yet.

[jdemeyer]

comment:17

Still: why fork cliquer instead of patching the sources?

[dimpase]

comment:18

Replying to @jdemeyer:

Still: why fork cliquer instead of patching the sources?

Cause this is the only way I know how to fix it once and for good. And cause this is waste of anyone's time to do manual building of shared libraries instead of using appropriate tools.

[dimpase]

comment:20

or do you mean to say "why don't I create a mega-patch instead of being honest about it?"

[jdemeyer]

comment:21

Replying to @dimpase:

or do you mean to say "why don't I create a mega-patch instead of being honest about it?"

If it is a mega-patch, you're probably doing it wrong.

[dimpase]

comment:22

Replying to @jdemeyer:

Replying to @dimpase:

or do you mean to say "why don't I create a mega-patch instead of being honest about it?"

If it is a mega-patch, you're probably doing it wrong.

Maybe you stop saying how wrong I am and review my changes?
This is getting silly. I made a sane directory layout to begin with, and created bog-standard Makefile.am and configure.ac

I forwarded you the email I just got from the maintainer. Go figure. I don't think we have this option in the "Report Upstream" menu.

[jdemeyer]

comment:23

Replying to @dimpase:

I made a sane directory layout to begin with, and created bog-standard Makefile.am and configure.ac

Which is very good, I'm not complaining about that. I am only saying that this is not a reason to fork cliquer.

[vbraun]

Reviewer: Volker Braun

[vbraun]

comment:24

Can you add a --libdir="$SAGE_LOCAL/lib" to the configure? I'm not 100% sure if its still needed now that we have a lib64->lib symlink; But some distros use a complicated multilib naming scheme and we'd still break if that gets used.

[vbraun]

Changed upstream from Reported upstream. No feedback yet. to Reported upstream. Developers deny it's a bug.

[sagetrac-git]

Branch pushed to git repo; I updated commit sha1. New commits:

8d7f039

added explicit libdir path

[sagetrac-git]

Changed commit from 3c17486 to 8d7f039

[dimpase]

Author: Dima Pasechnik

[dimpase]

comment:27

OK, done

[vbraun]

Changed branch from public/autocliquer to 8d7f039