This is from the Tuesday morning session, and lists features that the users would desire. A prioritised list of developers activities then follows.

• Datastream reshuffle (using Mark’s efficient code) to a standard VDIF always (kinda like interlaced VDIF - but maybe would rather do de-interlaced?)
• Beamformer inside DiFX. Have this be a separate data return path! Might even be easier just inside GPU-DiFX?
• Redoing the DiFX timestamping.
• Something about pysched? Helge will put in a ticket.
• Completing the initial GPU version!
• padVDIF has a bug - doesn’t work for single-threaded VDIF. Adam to check and fix.
• FiX the nsaccumulate floating point comparison error (Adam). Loosen by 10x??
• Update the VDIF mode name convention in configuration.cpp (Adam)
• Update difxcalc to avoid 36 station limit (Adam)
• Bug reporting pathway - will come naturally with git!
• publicise the DiFX slack space (and point to the DiFX repo's - git one, and Shaoguang svn2git one)
• VITA-49 support
• Separate pol colvert and HOPS from the DiFX repo in git
• DiFX 2.8.2 by the end of the week!
• Documentation of how to use the DiFX git repository (both as a user and as a developer)
• Natively support Mark6 recordings that span multiple modules? Some means to map a Mark6 module or group of modules to each v2d datastream, or to a VEX2 bitstream?
• Difx2mark4 in 2.6 and 2.8 has scaling issues with geodetic data? Simone for info?
• Difx2mark4 has another issue with amplitude scaling to fix, factor sqrt(2) too high, when duplicate autocorrelations are in mpifxcorr output; one auto for LSB recband, other auto for LSB flipped to USB zoomband
• Difx2mark4 multi datastream PCal support has made conversion extremely slow esp. compared to difx2fits, would be nice to fix this performance regression

Get all the DiFX contributors' git usernames! For Shaoguang.

To be filled in based on update of previous version

• Add complex data support to the GPU version
• Add LSB support to the GPU version - using unpack to flip sideband
• Calibrate the "absolute S/N" expectations for vdifsim
• Move the big difxcalc11 files into a separate "data" repository?? (and update difxcalc11 to then point to that extra repo??)