[Rivet] Rivet error

[Andy Buckley]

Hi again David, all,

I found the error -- just a stupid typo, in fact, where I rewrote a .pyx
suffix as .core rather than .cpp ... hence the error. I've pushed
updates to this to the repository: can you try it out. My build was
accidentally robust against it thanks to caching, and I'm not 100%
certain that it is fine now, so can you try it out and let me know.

I also added mappings of +, -, +=, and -= to YODA's Python wrapper this
evening, which were the last things on my "main" TODO list before
releasing a new version. The updated yodamerge script looks like it's
working -- here is the output from two small Rivet run YODA files being
merged together with the expected increase in bin height and reduction
in relative uncertainties:

http://users.hepforge.org/~buckley/yodamerge/MC_JETS/index.html

It is missing the sort of behaviour that Frank wanted w.r.t.
normalization, but this can either a) be added quickly, if we agree on
how to do it, or b) we issue Rivet 2.0.0 with a YODA version which can't
(yet) do that. I'm easy with either, as long as it's quick.

As far as I'm concerned all Rivet migrations have now been done,
validated in detail, YODA is in a "production quality" state (although
there is still so much to do...) and we can do things with histograms
that we never could before. So can we release Rivet 2.0.0 this week?
Assuming that everything builds fine for David, the rest of you (Macs?)
and on lxplus6, that is.

My plan after this release is that we rapidly merge the stockpile of
analyses that have been building up into both the AIDA branch and YODA
trunk versions of Rivet and issue 1.8.4 and 2.0.1 fairly soon. From then
on it's all version 2, and in particular we can work on getting jet
system overhauls and the funky Cuts stuff into 2.1. Sound like a plan?

Cheers!
Andy

PS. I'd still love to have a few Rivet hacking days somewhere this
autumn. Sorry for not following up on that, but I was busy and slots
kept getting booked up. And anyway, we wanted 2.0.0 to be out the door
first. Any good availabilities/venues at the end of this month or the
start of Nov?


On 01/10/13 12:02, Andy Buckley wrote:
> Hmm, thanks for finding that. I made a change last week which was meant
> to mean that people building YODA from a tarball (i.e. everyone except
> developers) would no longer need to have Cython on their system.
> 
> Due to the templated generation of .pyx files that Dave put in the YODA
> Cython build system, this was a more fiddly job than in Rivet or
> LHAPDF... so I probably screwed up somewhere but due to stale files it
> keeps functioning in my working area. I'll look into it, hopefully today
> along with other (final) YODA feature tweaks...
> 
> Andy
> 
> 
> On 01/10/13 11:33, David Grellscheid wrote:
>> Hi all,
>>
>> have you come across this build error recently?  I think it's something
>> to do with the latest Cython checkins, but I don't know enough in there
>> to fix it.
>>
>>> $ make V=1
>>> Making all in rivet
>>> make[1]: Entering directory `/mt/data1/herwig/src/Rivet-trunk/pyext/rivet'
>>> make[1]: Leaving directory `/mt/data1/herwig/src/Rivet-trunk/pyext/rivet'
>>> Making all in .
>>> make[1]: Entering directory `/mt/data1/herwig/src/Rivet-trunk/pyext'
>>> CC="clang" CXX="clang++" ARCHFLAGS="" /usr/bin/python setup.py install --install-lib=build/ --force
>>> running install
>>> running build
>>> running build_py
>>> running build_ext
>>> building 'rivet.core' extension
>>> error: unknown file type '.core' (from 'rivet/core.core')
>>> make[1]: *** [all-local] Error 1
>>> make[1]: Leaving directory `/mt/data1/herwig/src/Rivet-trunk/pyext'
>>> make: *** [all-recursive] Error 1
>>
>> Thanks,
>>
>>   David
>>
> 
> 


-- 
Dr Andy Buckley, Royal Society University Research Fellow
Particle Physics Expt Group, University of Glasgow / PH Dept, CERN