ROOT Team Meeting

Friday 4 Apr 2008, 11:00 → 13:00 Europe/Zurich

32/1-A24 (CERN)

11:00 → 11:05 News

-technical students; one requested by Bertrand for the event recorder -summer student bis; Magda could be with us in June/July (to be confirmed) -CMS event display demo yesterday -Rene seminar at Saclay next week (Thursday) -Bertrand poster at Swiss conference -Gerri to attend a GRID/PROOF meeting at Wisconsin -Fons to attend a Science&Technology conference in London (July 2) -new package roostats by Kyle Cranmer to be included in roofit/roostats (before end of April) -Maxime Moubarak will work with Fons as a summer internship

11:05 → 11:30 Progress Review

-Progress with removal of cernlib and import of a nanocernlib to build h2root (Olivier) -Progress with schema evolution (Philippe) -Progress with header files design for the multi-threaded CINT (Axel)

Slides api_proposal.pdf

From Olivier: A first pass has been done to extract all the needed routines in hbook zebra and kernlib. I am now trying to reduce this code to the minimum realy needed even by modifying the source code itself. The minimal hbook set of routines has been converted into hbook.f. Some routines have been changed to really load the minimal amount of code. The same work is going on for zebra. Zebra code is a bit more tricky than hbook. kernlib.f is going in parallel. Then there is also a bit of C. When the work will be finish (very soon I hope) we will have: hbook.f zebra.f kernlib.f some_c_code.c

11:30 → 11:50 Progress with SVN restructure (Fons)

-work done (side-effects) -to be done -fixes to get "make static" and the code checker working again

recently done or to be done net === alien auth glite globusauth krb5auth ldap monalisa net netx rpdutils srputils cintex --> cint/cintex ====================== core (done) ==== base cont meta metautils needelete pcre rint thread utils unix winnt fit, fitpanel ? =============== geom ==== gdml geom geombuilder geompainter graf2 ===== asimage freetype gpad graf postscript x11 x11ttf win32gdk graf3 ===== eve ftgl g3d gl x3d gui === ged gui guibuilder guihtml qt qtgsi qtroot sessionviewer io == castor chirp dcache gfal io mysql odbc oracle pgsql rfio sapdb sql xml xmlparser xrootd zip roofit ====== roofit roofitcore roostats (new dir to be introduced soon (Kyle Cranmer) <=== vmc ->montecarlo === eg g4root vmc pythia6 pythia8 pyroot, ruby --> cint ? ======================= table ->core ? ============

11:50 → 12:00 Documentation

Plans from team members to upgrade the classes/chapters documentation Class header description Parameters descriptions in a coherent/complete way Description of functions side-effects Any progress since last meeting?

12:00 → 12:10 Forum topics redesign

The Forum must be reorganized to take into account the FAQ. Must promote Roofit/Rootats and remove items rarely used CINT 3 items into 1 Replace support by bug-report and indicate the savannah url to report bugs

12:10 → 12:15 Web site redesign

Progress (Diego Casidei)

Diego Casidei is currently writing the top level pages. See current status at http://root.cern.ch/root/root-web The top page will be done (may be) by a company (Axel)

12:15 → 12:20 Article about ROOT in a special issue of Computer Physics communication

The following title has been submitted (thanks to all comments received): "ROOT: A C++ environment for petabyte data storage, statistical analysis and visualization" Some remarks were made about the title regarding "data storage"

12:20 → 12:40 Developments and Known problems

Development by Anar Manafov (in anar/branches) to add STL iterator-like to ROOT collections ======================================================================== Fons to investigate possible side-effects eve branches: please move to trunk =========================== Savanah bug reports (inflationary state). =============================== Those concerned to report at next meeting -Axel : 41 -pcanal: 27 -rdm: 15 -couet: 12 -moneta: 11 -ganis: 10 -bellenot: 9 The following problems reported By Eddy ================================ Problem1 (for Lorenzo) ======== I just ran into an annoying/dangerous feature of TGraph::Fit : The code below comes from TGraph::Fit if (linear){ // TClass *cl = TClass::GetClass("TLinearFitter"); if (isSet && strdiff!=0) { delete TVirtualFitter::GetFitter(); isSet = kFALSE; } if (!isSet && cl) { TVirtualFitter::SetFitter((TVirtualFitter *)cl->New()); It shows that if global fitter is set to TMinuit and a linear fit is requested, the fitter is deleted and replaced with TLinearFitter and that silently !! So if a user is performing non-linear fits and in between performs a TGraph::Fit("pol1") (aka linear fit) , he is toast in the next non-linear fit, unless he goes through the same code as above . If my analysis of the situation is correct, I request that TGraph::Fit restores he global fitter at the end of its operation . Problem2 (for Philippe) ======== I ran (again) into the following situation and wonder if it can be improved: If I run a script like: { TFile f("..."); TTree *t = (Tree *) f.Get("MyBranch") t->Process("MyCode+"); } and MyCode.C contains "TFitter", ACLiC of MyCode.C will fail in the link phase because (I guess) it uses the "root-config --libs" which does not contain -lMinuit . The same is true for any of my classes referenced in MyCode.C. I circumvent the problem by doing a t->GetEntry(0) which pulls in my libraries and a line with "TFitter fit;" in the script . However, I wonder if this can be resolved in a different way . Problem3 (for Axel) ======== Have a look at: http://root.cern.ch/root/html/TMath.html It does not look right, the summary overview of the functions ismissing. Scrolling down the list, I do not see for instance the Sort functions. Problem with HepMC1 to HepMC2 migration (Lars Sonnensheim + LHCb) =================================== Request to help for implementing custom streamers (no classdef, no version!) Request for option "width" in TProfile::Scale (Daniel Kramer) ================================= is the function in ROOT 5.19/02 TProfile::Scale(Double_t c1 = 1, Option_t* option = "") with option = "width" supposed to work like for TH1? I didn't work for me :o( Forum mail to be answered (Olivier) =========================== Multi TH2F on same pad http://root.cern.ch/phpBB2/viewtopic.php?t=6461