—————
2018/08/10
—————
Attendees: Philippe, Brian, Danilo, Oksana, Guilherme, Jim P, Taylor, Amit
Amit: presentation
Taylor: From here we want to extent the test to be more realistic maybe with Data from DUNE and lower sync frequency.
Brian: Fill cluster information during manual Flush (and also fix cluster iteration for ‘bad’ files)
Philippe: I think this is ready to merge.
Brian: Describing the ‘flush basket’ only at cluster boundary.
Philippe: It is a more complex PR, I will also check it today (with an eye on multi-threading issues).
Brian: I also been working on the bulk I/O branch.
Danilo: Back from holiday and focusing on ROOT User’s workshop agenda.
Guilherme: RAS.
Jim: Working on columnar manipulation that I will present at the workshop.
Oksana: I made some progress on LZ4 issues. We discovered that LZ4 is poorly compressing the offset array. (When disabling the storing of the offset array, the ‘problem’ goes away).
Oksana: before deciding what to do for the release, I want to test 2 possible work-arounds and will report on it next week [so we may not have go back to zlib as the default].
Oksana: I am also working on reducing dependency (including some circular dependencies and some ‘hidden’ dependencies) ; for example Tree depends on MathCore but CMake does not ‘know’ about it.