[Rivet] Phone/Skype meeting

[Andy Buckley]

Frank Siegert wrote:
> Andy Buckley, Thursday 01 October 2009:
>> ;) Yes, I saw that. I think while we still have to support earlier
>> HepMC versions (or do we?... discuss),
> 
> If ATLAS is using 2.05 now (you mentioned that in one of your last 
> emails?), I don't have any use-case for supporting older HepMC's if it's 
> problematic.

Agreed. I'm not exactly sure what ATLAS' position on this is: let's see 
if it's mentioned in tomorrow's MC meeting. It looked to me like after 
all the fuss in GENSER meetings about so much validation being required, 
suddenly it was upgraded one day without testing or announcement because 
LHCb wanted it and therefore GENSER had to start building everything 
else against 2.05. Not that I'm complaining, but of course I like to 
point out that I told them about 18 months ago that this was going to 
happen ;)

>> the detection needs to be done
>> in the C++ code rather than having to --- somehow --- build slightly
>> different Python interfaces for different HepMCs.
> 
> I am strongly in favour of this. Actually I'd go further and say that we 
> should probably move all the HepMC-related bits out of bin/rivet into the 
> analysis handler, it's only ~20 lines. If you think that's worthwhile and 
> possible, I could try that out. This would allow us to get rid of the 
> Python HepMC interface completely.

Yes, please try it out. Proud though I am of the crufty hacks which 
currently handle the Python interface for various HepMC versions in 
Rivet-HEAD, I'm *more* than happy to get rid of them. Let me know if you 
need any non-obvious hacks in pyext/rivet.i

>> We need to think about whether we can release Rivet 1.2.0 before
>> committing to a histogramming port: using better histo classes will
>> really open up what we can do, but it will be a rather major porting
>> exercise, involving changes to scripts which currently expect AIDA as
>> well as updating analyses which expect to use AIDA or LWH objects.
> 
> In my opinion 1.2.0 should come before the histogramming overhaul (which 
> you also called next-to-next release) and on a shorter timescale, i.e. O(1 
> month). As you said we have some actual improvements which we want to get 
> out before committing to an overhaul which probably will take some time 
> before the dust settles.

I completely agree. Good! Any other opinions?

Andy

-- 
The University of Edinburgh is a charitable body, registered in
Scotland, with registration number SC005336.