[Rivet] Rivet plans

[Andy Buckley]

Hi everyone,

It's been a while, and I know that there are a lot of people away in
various forms at the moment, but anyway it seems a good idea to get
planning in earnest for the next batch of Rivet development, working up
to release 1.1.3. How does a phone meeting next week sound?

The timescale I think we should be aiming at for 1.1.3 is the start of
May, nominally May 1st. That gives us a couple of weeks in which to deal
with the major issues on the tracker (happily, I've now worked around
the Python/SWIG bug that was looking like a serious blocker), and also
should allow us to migrate our jets to solely use FastJet behind the
scenes... provided that the FastJet release schedule itself doesn't
slip. Given the delay wih HepMC 2.05, we're going to have to wait one
more release before we can depend on the cross-section info being available.

The major issues in my sights are:

 * axis labels: I've added methods to specify axis x and y labels when
booking histos, but we need to migrate the analyses to use them. Once we
start this, I will be making the versions of the booking functions
without axis labels be quite verbose in their disapproval! This is
cosmetic but useful in the same way as the metadata was... it's a bit
embarrassing that our plots don't have axis labels unless we add them.

 * normalisations: Emily has overhauled the veto/sumOfWeights system,
since the idea of hiding event weight counters behind the scenes was
wishful thinking. We need to make sure that analyses are doing the right
thing, using explicit passed-cuts weight counters if necessary.

 * min bias analyses: not essential, but Atlas want to use Rivet for min
bias validation and we have a list of UA1/UA5 analyses that they want to
use. I have started work on this. Of course, there are many other
analyses that would be nice --- hopefully Hendrik will get some RHIC
analyses in as a consequence of his current Florida/BNL/FNAL trip ---
but as it approaches the deadline we'll have to prioritise on which ones
we really want to delay the release for. I do think it's important that
we increase the range of physics functionality with each release, rather
than just shuffle infrastructure... we're already doing well on that
score, largely thanks to Frank, but a bit more wouldn't hurt.

 * validator -> web pages: the validator has been a victim of a lack of
time, and the annoying complexity of trying to come up with a design
that combines the concepts of different Rivet versions, generators,
tunes, runs, and analyses without undue repetition. It is somewhat
unrelated to the release, but it would be good to be able to point to
web pages of gen/tune comparisons (cf. Peter Skands' plot pages, if
you've seen them, but for more generators) around the time of the
release. The degree of automation is, I think, debateable: we're not
really intending to cron this, at least not yet.

That's my brain dump for now... as I said way back at the start, I hope
we can have a meeting next week. Please let me know if that works for
you, and any other comments on the above.

Best wishes,
Andy