Bump Herwig to v7.2.0, ThePEG to v2.2.0

[mfasDa]

• Bump Herwig to v7.2.0
• Bump ThePEG to v2.2.0 (dependency of Herwig)
• Add CT14lo and CT14nlo to lhapdf-pdfsets,
  required by Herwig v7.2.0

[mfasDa]

Checking whether adding GMP as build requirement for ThePEG fixes the compiler issue of Herwig.

[marcovanleeuwen]

Hi Markus, @ktf,

It looks like the test-compilation above is not using defaults-generators. It is using fastjet v3.2.1 instead of v3.3.3. The -lgmp issue here is likely caused by that.

(fixing it here will not fix it for running on the grid yet; for that we need to fix the relocation issue)

Marco.

[qgp]

@ktf Fine with me, but has the problem of using the wrong defaults (pointed out by Marco in the previous comment #2153 (comment)) been addressed?

[mfasDa]

Could one trigger a new build test?

[mfasDa]

@maireiphc @preghenella I forced triggering the build test again and now it picked up the proper defaults, and all tests passed. I think one could proceed now.

Obsolete point (the alisw dir. of Herwig has been changed to the official one)

[mfasDa]

@qgp The build test is passing now - can the PR be approved now?

[preghenella]

hi @mfasDa ,
we have a issue with building AliGenerators in these days, that is failing in ThePEG.
I fear that if we bring this PR in, it might be harder to understand what is wrong.

Well, it would even be that this fixes it, but given that it appeared without any relation to a ThePEG upgrade, I would tend to wait a bit.

[preghenella]

@mfasDa , we got a successful build of AliGenerators this morning.
Thanks to @ktf for helping with that.

I think this PR can be merged.

[mfasDa]

Thanks a lot!

Once the PR is merged may I propose to think about decommissioning couple of generators which are shipped inside AliRoot in favour of linking against packages from AliGenerators. This affects PYTHIA(6,8), and LHAPDF. For the Herwig version (5) in AliRoot I am not sure it is currently in use but people should rather switch to Herwig7.

[preghenella]

Ciao @mfasDa ,
it would be possibly a good thing to do, but I am not sure how easy this is (will probably need to touch cmake in several places), who is supposed to decide on that step (Offline Board I guess) and who will actually execute the task...

[preghenella]

ciao @ktf , can we merge this PR?

I fear that we might encounter again the problem with YODA while building on Jenkins, given that we are upgrading ThePEG.
In case, we know how to deal with it... even though it is not an elegant way (we'll think about the right way to do it if indeed we hit the problem again).

Thanks,
Roberto

[preghenella]

that was perhaps expected.
@mfasDa , AliGenerators fails in building ThePEG.
https://alijenkins.cern.ch/job/DailyBuilds/job/DailyAliGenerators/354/console

The problems is that is seems that ThePEG does not find YODA, unless the YODA package is built in the same session.
I would say that there is a good reason to think that there is a issue in how ThePEG is pointing at YODA.
This conclusion comes from the fact that a recent PR in Rivet (that also uses YODA) did not show problems related to YODA.
So, Rivet is able to find YODA properly, ThePEG is not.

Do you have an idea what could be the problem?
I did not manage to reproduce it locally, this must have something to do with relocation matters.

We know that if we trigger a rebuild of YODA we can solve it.
But the problem will re-appear in the future, so I would be happy to find a solution forever.

Thanks,
Roberto

ps: @ktf , I know this is a hack, but is there a mechanism via alidist receipts that can be used to force a package build from another package? I mean, could ThePEG signal that he wants a YODA rebuild?

[ktf]

just add an extra space to yoda...

[ktf]

or a comment...

[preghenella]

uhm... maybe it is Rivet the problem and ThePEG relies on something that Rivet provides to locate YODA.
Will have a look

[preghenella]

what I meant @ktf was if one can do something in the receipt of ThePEG to force a rebuild on YODA.
Instead of the manual trigger via alidist update

[preghenella]

nevermind, for some reason today I can reproduce the error on my computer.
I think because I have downloaded YODA from alicache. let's see.

[ktf]

I think the correct thing to do is to understand what is happening, as it's effectively a bug. I suspect the problem is that we pick up YODA from rivet, rather than directly.

[preghenella]

I'm on it.
Somewhere while building ThePEG something is doing this grep and this sed

/usr/bin/grep: /mnt/mesos/sandbox/sandbox/jenkins/workspace/DailyBuilds/DailyAliGenerators/sw/20156152/1/slc7_x86-64/YODA/v1.8.2-2/lib/libYODA.la: No such file or directory
/usr/bin/sed: can't read /mnt/mesos/sandbox/sandbox/jenkins/workspace/DailyBuilds/DailyAliGenerators/sw/20156152/1/slc7_x86-64/YODA/v1.8.2-2/lib/libYODA.la: No such file or directory
libtool:   error: '/mnt/mesos/sandbox/sandbox/jenkins/workspace/DailyBuilds/DailyAliGenerators/sw/20156152/1/slc7_x86-64/YODA/v1.8.2-2/lib/libYODA.la' is not a valid libtool archive

but I cannot tell where.

Looks like explicitly adding YODA to the requirements of ThePEG does not do the trick.

[ktf]

Try grepping for YODA in Rivet. I think it caches some YODA path which gets out of sync.

[preghenella]

ps: I think ThePEG is able to correctly pick the correct YODA in general in this build.
In the Jenkins console you can find several instances where is works ok

unknown-linux-gnu/7.3.0/../../../../lib64 -L/lib/../lib64 -L/usr/lib/../lib64 -L/jenkins/workspace/DailyBuilds/DailyAliGenerators/sw/20156153/1/slc7_x86-64/pythia/v8243-alice1a-1/lib -L/jenkins/workspace/DailyBuilds/DailyAliGenerators/sw/20156153/1/slc7_x86-64/Rivet/3.1.1-alice1-1/lib -L/jenkins/workspace/DailyBuilds/DailyAliGenerators/sw/20156153/1/slc7_x86-64/fastjet/v3.3.3_1.042-alice1-24/lib -L/jenkins/workspace/DailyBuilds/DailyAliGenerators/sw/20156153/1/slc7_x86-64/cgal/4.6.3-34/lib -L/jenkins/workspace/DailyBuilds/DailyAliGenerators/sw/20156153/1/slc7_x86-64/MPFR/v3.1.3-63/lib -L/jenkins/workspace/DailyBuilds/DailyAliGenerators/sw/20156153/1/slc7_x86-64/GMP/v6.0.0-58/lib -L/jenkins/workspace/DailyBuilds/DailyAliGenerators/sw/20156153/1/slc7_x86-64/YODA/v1.8.2-2/lib -

see here, YODA is taken from build 20156153.

The line of the failure picks from build 20156152.

[preghenella]

I think I have found the reason for the failure, although I don't know how to handle the problem yet.

Looks like at some point ThePEG is using a libtool file from Rivet.

$RIVET_ROOT/lib/libRivet.la

I am not familiar with libtool but my feeling is that it looks into this file to determine the location of some dependencies.
And unfortunately, the file contains information about the absolute paths at the Rivet build.

dependency_libs=' -L/mnt/mesos/sandbox/sandbox/jenkins/workspace/DailyBuilds/DailyAliGenerators/sw/20156152/1/slc7_x86-64/YODA/v1.8.2-2/lib -L/mnt/mesos/sandbox/sandbox/jenkins/workspace/DailyBuilds/DailyAliGenerators/sw/20156152/1/slc7_x86-64/HepMC/HEPMC_02_06_10-5/lib -L/mnt/mesos/sandbox/sandbox/jenkins/workspace/DailyBuilds/DailyAliGenerators/sw/20156152/1/slc7_x86-64/MPFR/v3.1.3-63/lib -L/mnt/mesos/sandbox/sandbox/jenkins/workspace/DailyBuilds/DailyAliGenerators/sw/20156152/1/slc7_x86-64/GMP/v6.0.0-58/lib -L/mnt/mesos/sandbox/sandbox/jenkins/workspace/DailyBuilds/DailyAliGenerators/sw/20156152/1/slc7_x86-64/cgal/4.6.3-34/lib -L/mnt/mesos/sandbox/sandbox/jenkins/workspace/DailyBuilds/DailyAliGenerators/sw/20156152/1/slc7_x86-64/boost/v1.72.0-alice1-37/lib -L/usr/lib -lboost_thread -lboost_system /mnt/mesos/sandbox/sandbox/jenkins/workspace/DailyBuilds/DailyAliGenerators/sw/20156152/1/slc7_x86-64/YODA/v1.8.2-2/lib/libYODA.la -ldl -L/mnt/mesos/sandbox/sandbox/jenkins/workspace/DailyBuilds/DailyAliGenerators/sw/20156152/1/slc7_x86-64/fastjet/v3.3.3_1.042-alice1-24/lib -lfastjet -lCGAL -lgmp -L/jenkins/workspace/DailyBuilds/DailyAliGenerators/sw/20156144/1/slc7_x86-64/cgal/4.6.3-34/lib -L/jenkins/workspace/DailyBuilds/DailyAliGenerators/sw/20156144/1/slc7_x86-64/GMP/v6.0.0-58/lib -lm -lfastjetplugins -lsiscone_spherical -lsiscone -lfastjetcontribfragile -lfastjettools -lHepMC -lz'

If I manually edit the the line

/mnt/mesos/sandbox/sandbox/jenkins/workspace/DailyBuilds/DailyAliGenerators/sw/20156152/1/slc7_x86-64/YODA/v1.8.2-2/lib/libYODA.la

which is the line where ThePEG complains about, with the correct path on my system, I can make ThePEG compile.

I will have to find out how things can be properly done when relocating.
If you have hints where to put hands, thanks and I will have a look.

Roberto