Present: Lynn Garren, Herbert Greenlee, Alex Himmel, Thomas Junk, Marc Paterno, Gianluca Petrillo, Ruth Pordes, Brian Rebel, Saba Sehrish, Erica Snider, Tracy Usher, Karl Warburton
Connected: Vito Di Benedetto, Vyacheslav Galimov
Status of the project [Erica Snider]
-------------------------------------
- upgrade develop to LArSoft v5 is proposed for the beginning of next week
- MicroBooNE MCC7 will be given hot fixes on the production version (v4)
as needed, as usual
- GENIE 2.10.4 is going to be adopted by NOvA
- [Lynn Garren] GENIE 2.10 should go first as the change seems seamless
- [Lynn Garren] testing can be helped by Continuous Integration system
- [Alex Himmel] validation was seamless; can turn on meson exchange and other models
- no objections to have GENIE 2.10 go in first
- the next upgrades will be GEANT4, then ROOT 6, in that order
- art is already using ROOT6, depending on changes it may result in a major release
- art is being split as the result of LArSoft/LArLite integration sub-project
- this will affect LArSoft as soon as we go to art 1.18 or higher (ROOT 6 is required)
- OS survey showed the most used ones to be Ubuntu and then OSX 10.11 ("El Capitan")
- 47 responses in total, 23 would like to use Ubuntu
Conclusions:
- stakeholders will be informed by e-mail of the intention to start merging
LArSoft v5 to develop branch, and will have the chance to object
- GENIE 2.10 will be the first of the upgrades
Continuous Integration Operations [Vito Di Benedetto]
------------------------------------------------------
- C.I. running on SLF6 and Darwin 13.4 (OSX "Maverick")
- [Lynn Garren] we should be running on Darwin 14 (OSX "Yosemite")
- more experiments are going to be added
- we have plots of resource usage history
- the adoption of CORSIKA triggered a doubling of CPU usage and significant increase in memory (2GB to 3.5 GB) in MicroBooNE tests;
MicroBooNE has been informed
- MicroBooNE redo stage 1 is not reproducible
- OSX test using IFDH ls command is failing because that command is not supported
- [Lynn Garren] a workaround is supposed to have already been in place
- [Lynn Garren] a new release for IFDH will support natively a workaround
- it will provide a directory listing file that can be read in
Conclusions:
- run tests on OSX 10.10 (Yosemite)
- check why the IFDH workaround is not in place as it was expected
Proposed change to ExternalTrigger data product [Karl Warburton]
-----------------------------------------------------------------
- time stamp for trigger was needed to be larger than 32-bit
- issues with ROOT labelling differently uint64_t in OSX and Linux
- [Marc Paterno]
- Doxygen is not showing the changes
- [Lynn Garren] that's because it picks the higher tag, that is v5
Conclusions:
- changes are accepted (they are already in)
Any other business
-------------------
[Herbert Greenlee]
- opening the topic of a change in a data product changing a member from short to int
- in LArIAT there is DAQ header data product, this is added to an event and then never looked at. It has a 16-bit number for event number, the proposal is to use 32-bit.
- will discuss more extensively in the coming weeks
- will prepare a feature branch
LArSoft simulation of the Dune 10kt Dual-phase TPC [Vyacheslav Galimov]
------------------------------------------------------------------------
- Completed with horizontal drift but missing lights readout: optical detectors, all available in v_04_33_00
- DetSim updates
- There were some warnings as a result of new changes in the simulation
- Event display: not managed to display the raw data, it crashes
- Do we understand the crash? It hasn’t been investigated yet.
- [Tom Junk] non-linearity can be added after the convolution
— What happens to charge in the gaps?
- there is always a wire nearby to collect the charge
Optical Hit reconstruction [Kazu]
------------------------------------------------------------------------
- Change made for MicroBooNE data reconstruction
- larana/OpticalDetector
- Added the ability to configure algorithms
- the changes Kazu has made in form of new configurable algorithms
- Gleb is working on separating Flash finding and hit finding
- Kazu and Alex will coordinate if there are any changes needed in fcl configuration for dunetpc
Hit Finder [Tracy]
------------------------------------------------------------------------
- proposed changes to hid finder
- three main functions should be broken down into independent functions
- single function to do gaussian
- question to tracy: will you take out the code and break that into algorithm and module.
- approved: ready to go into develop
Changes to radiological and supernova generators [Gleb]
------------------------------------------------------------------------
- couple of features fhicl parameters were added
- in larsim: feature/gvsinev_SNGeneratorImprovements
- Feature branch in LArSim: feature/gvsinev_Rn222Background
- Feature branch in LArAna: feature/gvsinev_PerEventFlashTree
- a new tree with flash information stored for each event, off by default
- all ready to be merged - in this release
LArSoft build speed issues [Marc Paterno]
------------------------------------------------------------------------
- in a particular change form v_04_26 to 29, build time was increased significantly
- change in fhicl cpp that caused the change, there was a template being used in an inline function causing that template to instantiate at many many places,
- move the code from inline to the cc file
- you should really inline if you need to,
- library sizes came back to normal size and times are back to normal
- Tingjun
------------------------------------------------------------------------
- something approved last time was not ready, but will be ready in few hours.
- It will be merged in the next release