ROOT Team Meeting

Friday 7 Mar 2008, 11:00 → 13:00 Europe/Zurich

32/1-A24 (CERN)

11:00 → 11:05 News

Diego MAPS ACAT08 Erice early November Next ROOT workshop in Saas-Fe (Eddy) Matevz success with CMS event display (must be ready for April 1st) good news Summer students: happy winners, send me the names and functions Summer students with Axel ===================== here are "my" summies: Wieslaw PALKA will work on the automatic, on-the-fly generation of dictionaries via a callback from CINT to ROOT (which then invokes rootcint etc). Gencer ERDOGAN will work on the THtml improvements: using Reflex, documenting more entities, structuring my namespaces etc - and probably also function parameters / return types. Summer student with Bertrand: ======================== OPOCENSKA Katerina The candidate will be working for the ROOT project in the Graphical User Interface (GUI) work package to participate in the development of Quality Assurance (QA) for the GUI. The assignment of the candidate will be to investigate and evaluate existing GUI validation tools available on Linux and Windows. If any of these tools is suitable for the ROOT GUI, then his/her work will concentrate in the integration of the selected tool with the GUI for QA and stress testing. If no tool is found, he will be developing a GUI event recorder/player to be used in the validation and stress tests of the GUI, as well as for an automated player to be used mainly for tutorial purposes.

11:05 → 11:30 Status of dev release 5.19/02 scheduled for March 15

Each work package will describe its readiness status. In particular -situation with the new libMathCore -new CINt/Reflex merge and build structure -Will libAfterImage be expanded in SVN like the new xrootd?

Slides CINTDevMarch08.pdf CINTDevMarch08.ppt

From Axel ======== see Axel's ppt file From Wouter ========= RooFit status in short for 5.19 - Found & fixed a couple of small but important bugs that should go into 5.19/02. Most are in already the very last ones are only in my dev branch. I will aim to get the trunk ready by tomorrow. - There is ongoing development work on the RooWorkspace class for 5.19, but it is not clear that this is in sufficiently good shape yet to be released. - Some additional developments planned for RooStats support. Will not detail here unless you ask. - Most of the coding work on RooFit for the June prod release should happen in March/April.

11:30 → 11:50 SVN structure with a hierarchy reflecting more the projects

When do we implement the new structure decided at the previous meeting. Proposal to do it immediatly after 5.19/02 release

11:50 → 12:10 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

12:10 → 12:20 Web site redesign

-News (Axel) about the cosmetics/graphics of the main pages -Possible contribution to write an introduction to the main topics

12:20 → 12:30 Milestones

Comments for quaterly report and proposal for new milestones

During the last quarter, we focused mainly on the quality assurance procedure to deliver a good production release 5.18 in January. This new version includes several new packages and consolidation of existing packages. The release notes are visible at http://root.cern.ch/root/Version518.news.html. The QA procedure called roottest consists of about 150 major tests, each one including between a few and 50 specialized tests with a total of about 35000 lines of code or makefiles calling in turn 60000 lines of code of compiled or interpreted tutorials. The result of the nighly builds is visible at http://lcgapp.cern.ch/spi/aaLibrarian/nightlies/ Comments on existing milestones ================================ ROOT-16 ======= >30.06.07 31.12.07 >Cint 7.2 will use Reflex for storing all information regarding types (aka replace the G__struct global array). >In progress. >Rescheduled. We had successfully merged Reflex into CINT in December 2007: CINT now uses all of Reflex to store its reflection data. This new version of CINT cannot be put into production: Reflex makes it almost 20 times slower than the original version of CINT, and several prerequisite features (delayed loading of dictionaries, using directives etc) are missing in Reflex. Reflex is currently rewritten to address all of these issues. We plan to introduce the new CINT as an option into ROOT in the March development release. We plan to switch to the new CINT as default for the December production release of ROOT. ROOT-17 ======= >30.06.07 31.12.07 >A major upgrade of the 2-d graphics system calling directly OpenGL is under investigation. If we succeed: >- one we could get rid of the X11 and win32 interfaces >- speed-up the graphics >- easily mix 2-d and 3-d graphics in the same OpenGL viewer >A report on this work will be given at the ROOT workshop end of march. >One could hope to have a full implementation by end 2007. >In progress. >Rescheduled. A Canvas can contain pads with OpenGL 3D graphics rendering. It means that in the same graphics window some parts are painted using OpenGL (detectors geometries, Lego plots, Surface plots, etc ...) and other parts are painted using basic X11 2D graphics. PostScript files generated from such "mixed" canvases are a combination of gl2ps output (for the OpenGL parts) and normal ROOT PostScript output (for the X11 parts). The possibility to render 3D text A(using FTGL) and 3D axis has also been implemented. Despite the fact it has been now demonstrated that it's possible to mix standard 2D graphics and 3D OpenGL graphics in the same pad, the complete replacement of X11 and win32 by OpenGL has not been completed. One of the main reason was that the memory overhead introduced by OpenGL (25M) was bigger than ROOT itself. The developments around GL are now more focused on the Events Display with the new EVE (Event Visualization Environment) package. ROOT-18 ======= >31.12.07 >Specifications of the API for the support of complex data schema evolution >in ROOT >New The planned extensions to the existing class schema eveolution system in ROOT have been discussed with Atlas and CMS during the Autumn and presentations of the API indicating the way it will be supported made to both collaborations in September and October. It was agreed that the new system should be in the June release. A more detailed design is currently happening (collaboration with Atlas). We still expect to have the new system in production for the June release. Proposed new milestones ======================== ROOT-XX 30-06-08 ======= Test suite improvements: In the past few months a major effort has been invested in the QA procedures. Several hundred tests have been developed and run every day in the nightly builds. We want to consolidate this test suite to improve the robustness of the system in general. We also need to invest more time in testing the new (or development) versions of the compilers on all systems (eg today gcc4.3 or VC++9). This requires that the SPI team provides access to these new compilers in the nightly build system. ROOT-YY 31-12-08 ======= Restructuring of the web site and documentation system: We are planning to make a substantial reorganization of the ROOT web site (unchanged since many years) to reflect a more modern presentation style. The new site will include a brief description of the main ROOT functionalities and more guidance for newcomers. The class documentation system will be revisited to provide a more coherent description of the class, function parameters and side-effects. A first version of the web site is expected for June 30. ROOT-ZZ 30-06-08 ======= Event Display library. A big effort is currently going on to provide a powerful experiment independent library suitable for high performance event displays. We are currently collaborating with Alice and CMS for the development of their event displays. Both are expected to be in production this spring. Several people in Atlas have also expressed a big interest in this library. Although more developments are expected in this area, the two prototypes for Alice and CMS should give good indications of our directions. ROOT-PP 31-12-08 ======= PROOF: Implementation of PROOF optimized to run locally on multi-core platforms. This version of PROOF, PROOF-lite, will not use the xrootd daemons, but start directly the master and workers. Communication will be via local mechanism, like Unix sockets and message queues. Also this version will optimized the I/O by trying to use memory mapped I/O. Due date December 2008.

12:30 → 12:35 Article about ROOT in a special issue of Computer Physics communication

Dear Rene, *40 years of Computer Physics Communications * In 2009 Computer Physics Communications (CPC) reaches its 40th anniversary. With over 2,000 computer programs contributed since 1969 this unique journal (http://www.elsevier.com/locate/cpc/) and its associated Program Library (http://cpc.cs.qub.ac.uk) is a major computational resource for the scientific community. To celebrate this event we are planning a thematic issue focused on quality software for high performance, grid and novel computing architectures. The theme will be interpreted broadly and is intended to include high end computing using supercomputers, parallel computing using clusters, codes that exploit the cache architecture on a single node, novel architectures such as FPGAs, GP GPUs, Cell processors and novel computing techniques such as genetic programming. Software that exploits strategies for parallel computing, large-scale computing, and algorithms designed to carry out difficult scientific calculations efficiently is particularly welcome. /On behalf of CPC's Editorial Board we invite you, and members of your team, to contribute to this special issue by publishing one of your important computer programs./* *Contributions are by invitation only and will be subject to the normal peer review process. Submissions are required for peer review before *31 March 2009*, with publication planned for *November 2009*. There will be a page limit of approximately 30 journal pages per article. Abbreviated author instructions for this special issue, and a LaTeX template, can be found at http://cpc.cs.qub.ac.uk/CPC_40_Years.html. The full CPC Guide for Authors can be found at _http://www.elsevier.com/locate/cpc/ _. FAQs on program submissions can also be found at the Library's home page http://cpc.cs.qub.ac.uk. CPC normally publishes two kinds of paper: regular computational physics papers and papers that describe programs submitted to the Program Library. Early issues of the journal only published the latter. Therefore, in keeping with the historical, celebratory theme of this special issue we have decided that it will only contain program descriptions. Please note that: * The copyright and all other rights remain with the program's author. By submitting a program to the Program Library the author gives the Library permission to distribute it to those eligible to receive it under the Program Library's distribution service. * Library programs are free to members of institutions that subscribe to the CPC journal (print or electronic). Individuals who do not belong to subscribing institutes may take out an annual Library subscription. * All programs are distributed under CPC's standard licence (http://cpc.cs.qub.ac.uk/licence/licence.html). However, if you wish, you may provide the library with a licence tailored to your own requirements. * The Library e-mails authors each quarter with statistics on the number of downloads and, if available, the e-mail addresses of the individuals who downloaded. To promote your work and that of the journal the publisher has agreed to provide the following special issue features. * The on-line version of the thematic issue will be free to anyone for the first four months after publication. * A free copy of the entire thematic issue will be provided for every co-author. * A free copy of the entire thematic issue will be provided for 6 leaders in the field nominated by the principal author. * The thematic issue status will be highlighted on the front cover of the print journal and above the manuscript title of each pdf file. Thus authors reading other articles in this issue will be aware that it is a thematic issue, and this will encourage browsing of and awareness of the whole issue. * An e-mail-shot will be made to CPC readers announcing the thematic issue and providing a link to it. * The journal website will highlight the thematic issue and provide a link to it. To help us plan for this special issue I should be grateful if you would let me (ns.scott@qub.ac.uk <mailto:ns.scott@qub.ac.uk>) know by *28th March, 2008* whether or not you accept this invitation. If you do accept please provide a tentative article title and program name. Suggestions of colleagues who you think should contribute are also welcome. We look forward to hearing from you. Yours sincerely, Zbigniew Was, Principal Editor HEP, CPC, (zbwcpc@cern.ch) Stan Scott, Coordinating Principal Editor, CPC, (ns.scott@qub.ac.uk) Don Truhlar, Principal Editor, CPC, (truhlar@umn.edu) Peter Coveney, Specialist Editor (HPC, Grid and Novel Computing), CPC, (P.V.Coveney@ucl.ac.uk)