ROOT Weekly Meeting

Friday 8 Jul 2005, 11:00 → 12:15 Europe/Zurich

32/1-A24 (CERN)

Rene Brun (CERN)

11:00 → 12:15 Tour de Table

Rene ==== -selection board for the "8th SFT post" took place on Monday 3 candidates. Only 2 came. Axel selected. Could start in October. -Valeriy Onuchin departure. Valeriy will (very likely) continue to work closely with us from Protvino (may be with Alice) -ROOT version 5.02 development release seems to be OK. -Bug fix release 4.04.02d applied this week. The list of bug fixes is shown at: http://root.cern.ch/root/Version40402-patches.news.html -request to apply another bug fix in 4.04/02 by Giulio Eulisse to run in multithreaded mode with ROOT/Qt. -Next development release planned around August 15 and next one just before the workshop end of September. -sundev has disappeared. No way to test Solaris version. -Workshop page now available from the ROOT main page. Using the new Indico Conference/Workshop management system. Please register. -Philippe/Masa making progress with CINT version 6. Philippe making the necessary changes to compile CINT with C++ -We are in the process to collect data and selectors for the PROOF demo at the workshop. We will have data from Phobos and alice (in progress). Hope to get new data sets from CMS end of August (Craig Munro from ARDA following the PROOF weekly meetings). -Google seminar yesterday. Happy to see convergence of ideas. got some new ideas. -Timur Pocheptsov will be with us for 2 weeks (18 July). -Many complaints about Savannah. Hope that our complaints will be taken into account soon to improve the interface. -I will be on holidays next week. Markus ====== Patch release from last week seems to work. Corresponding POOL release will come next week. Pere ===== Waiting from Fons for the proposal for the plugin-manager --------------------------------------------------------- Used latest version of PyRoot. Works, except for STL vectors, list, etc strange problems related to the fact that CINT defines dictionaries for STL containers. to be checked with Philippe. Guenter ======= I currently extend xrootd/olbd for the auto discovery of services using apple bonjour. So in a local and wide area network the xrootd's should be able to find each other automatically. implementation advances: on the local link, the olbd dataservers are able to find the olbd master(s) and connect to them. code for service discovery using apple bonjour is in an external library which is loaded dynamically at runtime. so olbd does not rely on bonjour and can run and compile without it. other points: apple bonjour zero-config is supported for example in Fedora 4. Fedora 4 includes howl <http://www.porchdogsoft.com/products/howl/> which is an implementation of apple bonjour. however the client API is different to the one provided by Apple. I contacted the guys from Howl to tell me other differences, for example if a howl client can work together with the apple mDNSResponder or vice versa, communication with a DNS server... Waiting for an answer right now. apple bonjour for wide area discovery: in networks where the machines are separated by routers multicasting does not work necessarily. bonjour uses a "trick" by sending a DNS Update query to the DNS server. This has to be supported by the DNS server. I talked to Nick Garfield from IT if it would be possible to use the CERN DNS server for this purpose. It's not as easy, but he could provide us a domain after an official request and clarifying some open questions. For testing purposes he will help me with the configuration of a local DNS server on my local machine and afterwards provide me an unofficial test domain on the CERN DNS server. Additionally he recommended me to contact Vlado Bahyl who is also dealing with service discovery but using UDDI (this I will do of course). I also contacted the developer group of Apple regarding the communication between the mDNSResponder and the DNS server using shared secrets for update request. For Apple everything works fine, for Linux a slight modification has to be done to get the shared secret from the configuration file. (this means to change the code in a third party software and hence using a non-standard version!!!) Maybe it would be good to implement and provide more than one technology for automatic service discovery. this I have to think about... next steps: - clarifying the points mentioned above - look how we can integrate olbd into xproofd Jose ==== I've been working on the indentation program. I managed to indent C++ keywords, such as, public, private, according to Root convention. Still some test to do. Bruno ===== Used this week to test and compare results on the Qt layer of root, comparing it to X11. Several bugs and not documented features and limitations of the Qt layer where identified and communicated to Valeri Fine. Some corrections derived from those exchanges with Valeri. Andrei,Mihaela ============== - Possibility to work with ROOT geometry for G3-based simulation of ALICE either when this is created via VMC calls or when loaded from file. Simulation results identical in the 2 cases after some cleanup was completed in AliRoot. - updates in TGeo documentation - some preliminary work for the geometry interface G4-TGeo Andras ====== The calling of Mathematica functions from within ROOT now works, i.e. one can assign a Mathematica function to TF1 and display it. The probability density functions and their cumulative counterparts checked up till now show a relative difference of approximately 10e-15 between Mathematica and MathCore. Olivier ======= * Pattern in PDF files are done. As some of the patterns in PostScript are build with little programs (which is not possible in PDF), I converted with acrobat distiller the PS file in PDF, then I extracted the zip compress part of the pattern description (thanks to the "od" command) and put it in the PDF pattern definition. I had to redo some patterns because they were not visible (lines too thin) when printed. Note: Pattern (1 to 25) are still missing in SVG but we had not urgent request yet. Should I spend some time on it ? * Some more work on TPie (not yet finish). * Trilinear interpolation in TF3::GetSave * Patch for 4.04/02 (TH3::Project). In some cases errors bars were not correct in that version. * Some user (Karina.Loureiro@cern.ch) reported me a bug on Mac. I succeeded to reduce her problem to something very small: loop () { Long64_t n=10; Long64_t i; for (i=0; i<n; i++) { printf("%d\n",i); } } On Mac the loop is done only once. If I replace Long64_t by Int_t it works. I asked her to forward this problem to roottalk but I an not sure she did. Ilka ==== Working on the user's Guide updates: - geometry chapter (Mihaela) - cint and tree chapters (Philippe) - gui chapter - qt interface added (Valeri) * waiting for Python chapter update from Wim Testing Valeriy's patch with the image plugins that allow to display images of several formats in a canvas. In addition, PS/EPS/PDF and text files can be browsed with the root browser. Anna ==== continue working on the loess algorithm, especially on multi-dimensional interpolation of a surface, based on boundary curves. Studied different such interpolation methods and probably found one appropriate for the problem. by the way, if you are interested, here is a good description of my target program: http://www.itl.nist.gov/div898/handbook/pmd/section1/pmd144.htm Gerri ===== In the last week I have worked mostly to the reorganization of the current PROOF code to prepare the integration of the new components. I have additionally cleaned-up TSlave and TProofServ by separating out the current authentication code from LibCore and LibProof into a plugin loaded by TSocket only when needed. I've also prepared the patch for the new version of TXNetFile which I will submit to Fons later today. This contains new authentication plugins for xrootd which will then be activated in XPROOFD.