Telco 20170711¶
Date¶
Tuesday, 11 July, 16:30 CEST (+0200)
Hangout Link: https://plus.google.com/hangouts/_/j72qwlvegiojjpt3a36pfhow5ua
Agenda¶
Welcome and agree Agenda
Code Camp or similar activity in 2017
Update on current projects
ready to release v3.3 of NeXus definitions
Adding PDB tags to NeXus for NXmx (as per Lund HDRMX meeting)
NXcontainer - look at example files in pull request to examplefile repo
identifying data source location in NXdata
NXslit proposal
Cspad to NeXus: NXdetector_module
NXreflections - ragged arrays?
Github HDF5 filters repo
Next meeting
AOB
Minutes¶
Present: MB, MK, PJ, AB, TSR, EW, FA, HJB
Code Camp 2017¶
Target is last week of October, for about three days, hosted by DLS.
Ongoing Projects¶
Release 3.3¶
Looks like it is ready to go. There is an automated script that generates release notes. Mid term they could move into the wiki part of that repo space.
Target date for the next release is January, or maybe earlier at the Code Camp?
PDB¶
All are in agreement that this is a good idea. HJB talks to COMCIFS and COMDATA around the IUCr meeting in August. We can look at the implementation at the code camp.
NXcontainer¶
Nothing to report and on the back burner. DLS suggested dropping from the agenda until some progress is made there.
Identifying data sources in NXdata¶
DLS have prepared and example and no one could come up with a better proposal. They will put this to the test in practise now.
NXslit¶
We clarified the preferences raised in emails and the ticket system with the conclusion that “left” and “right” are unnecessarily vague concepts possibly overloaded with local conventions. Positive/negativ x/y direction is directly tied to the NeXus coordinate systen and will be the choice for the new proposal.
CSPAD¶
We did not discuss the CSPAD, but aligning timestamped data in NXlogs written with different frequencies using the cue indices.
AB would like to see some full example file. TSR may have one or could create one for one of the next meetings. MB indicated the same. Would be nice to compare.
Ragged arrays¶
They are more of an HDF5 problem than a NeXus one. There is no simple efficient way to deal with them. Either you end up with many unused chunks and lots of metadata or you need to flatten the dimensions and have a run length index dataset. AB will investigate relative merits with LCLS people.
HDF5 filters repo¶
Eugen is making progress on simplifying the build system, which is currently too complex to maintain.
Next Meeting¶
AOB¶
Dropped due to lack of time.
