Choose timezone
Your profile timezone:
Powered by Indico
v3.3.3-pre
Present: Charis, Lazaros, Sylvain, Barth, Vasco, Adam, Markus, Ruben, Giulio, Pierre, Dario, Peter
Remote: Alexey, Mohammad
1. Documentation
Barth: I tried an external project, somehow different from the tutorial. It is sending and receiving histograms. I had to implement CMake build system myself and to extend the FindFairRoot macro.
Mohammad: Now we have possibility to specify the version of the external packages using a configuration file.
Barth: What is the purpose of the state machine?
Mohammad: It is needed in case of many devices to guarantee the correct call chain. For example, you send data only when the devices are ready. We already discussed the requirements from the control system and the current implementation takes them into account.
Barth: There is also the #ifdef nanomsg/0MQ question...
Mohammad: It is a remnant of the past that will be removed. Now we have possibility to switch between nanomsg/0MQ without recompilation. Please do not mix the two transport engines!
Giulio: It is probably interesting to have this private mail exchange posted to the e-group.
Peter: We also discussed the possibility to use dedicated forum for information exchange. Of course, we can still use the e-group, but the forum is more visible for outside people.
Sylvain: The interesting discussions and topics should be migrated to the documentation.
Alexey: The GitHub issues are good alternative to the forum. It is possible to get notification by e-mail.
Barth: We also should create dedicated JIRA for the software project (separated from CWG13).
Giulio: Since we use GitHub, the GitHub tickets are also alternative to JIRA.
Pierre: We should ask CWG11 to report the results of a comparison between GitHub tickets and JIRA. Also we should remind the choice of GitLab vs GitHub. Is GitLab accessible for the outside people.
Dario: CWG11 already reported on the GitHub/GitLab subject and we agreed on a decision. We will provide information about the tickets vs JIRA.
Peter: I have a question about the AliceO2 Web page. How should we use it?
Barth: We should move from CWGs to software project. Jeremi will make a presentation about the ways to update this page.
Peter: Lets move to the AOB. Last time Lazaros presented his plan for 0MQ RDMA implementation. After three weeks of tests we decided that it would be much easier to use nanomsg.
Pierre: There is related topic. We had an interesting presentation at CWG3 meeting about the network performance of the AliceO2 topology. Using 40Gb Ethernet we see we need only one core for the data transport. Adam will present updated results next week at the CWG13 meeting.
Mohammad: Please use also the standard 0MQ benchmark test to be able to compare with the old results.
Peter: The last AOB is about Root6. The memory overhead in the present AliRoot reconstruction comes from a misuse of the Root plugin manager that creates the reconstructors and the loaders for each detector.
Markus: We should use factory class to solve this issue.