User Tools

Site Tools


difx:difx-todolist-2012

Differences

This shows you the differences between two versions of the page.

Link to this comparison view

difx:difx-todolist-2012 [2015/10/21 10:08] (current)
Line 1: Line 1:
  
 +===== To-Do and Wish List =====
 +
 +== High Priority ==
 +  * <​del>​Cluster configuration file standard definition
 +    * Adapting various clients (genmachines / threads) to make use of cluster definition file</​del>​
 +  * <​del>​Add large (~1 TB) dataset with variety of modes on Data Fabric (CR)</​del>​
 +    * <​del>​Synchronize data fabric test datasets at several sites. (Bonn, Curtin, NRAO) </​del>​
 +    * Regression testing scripts (AD)
 +  * <​del>​Amplitude correction for primary beam effects for multifield projects (available in AIPS now, but not fully tested)</​del>​
 +  * <​del>​IPP alternatives</​del>​
 +  * Xcube support (at least one way of getting Xcube data into DiFX)
 +
 +== Medium Priority ==
 +  * <​del>​A watchdog which can kill an unresponsive job after a configurable amount of time (HR) </​del>​
 +  * Timerange selection in difx2fits
 +  * <​del>​Speed up interlaced VDIF code for nativemk5datastream </​del>​
 +  * Fix the problem where interlaced VDIF packets don't all start at the same time for all threads
 +  * Find and fix the millions of MODE errors that sometimes occur (ATD) ** Thought to be fixed **
 +  * Add project code to difx messages
 +  * Choose framework for unit testing. Implement test cases (JM)
 +  * Add DiFX diagnostics
 +  * <​del>​Storing the original pointing source somewhere in the FITS file (plus the attendant AIPS changes)</​del>​
 +  * Add per-IF phase polynomial in the calc file (and delay polynomial, too?)
 +  * Documentation
 +    * Doxygen of mpifxcorr source - <​del>​automatically installed locally</​del>,​ also someone mirror to a webserver
 +  * evlbi
 +    * reconnection/​long term drop-outs (CP)
 +    * reliability with UDP data (CP)
 +    * real-time delay/rate adjustment (CP)
 +  * Maintain database of benchmarking results (JM)
 +  * Phased array (RS, CR). 
 +  * visplot (CP, JS)
 +  * Investigate sampler stats monitoring (AD/WB/CP)
 +  * subband distribution/​parallelisation (with VDIF) (AD/CP)
 +    * replumbing/​refactoring vex2difx setup
 +  * Transition to using band centres instead of band edges (WB)
 +  * Baseband data generator for producting test datasets. (WB)
 +  * Parallelised calc server / Efficiency gains in CALC (same time, different sources etc)
 +  * Store polynomial tau(l, m) generalized delay model in FITS (WB,JM)
 +  * Chase up and eliminate the cause of occasional "​Somehow Visibility X ended up with Y+1 subintegrations - was expecting only Y" errors ** Thought to be fixed **
 +  * Mixed module and file list on one station
 +  * Convert all LSB bands to USB on unpacking
 +
 +== Low Priority ==
 +  * Check the relative signs of 1-bit / 2-bit unpacking
 +  * Source selection / deselection in difx2fits
 +  * IF selection in difx2fits
 +  * addZoomFreq in the SETUP section (applies to all antennas)
 +  * Documentation Algorithms
 +    * monitoring, etc - (list of error codes) (AD)
 +  * Play back from both Mk5 banks in parallel (WB)
 +  * Still produce one FITS file even when there was a clock change during the correlation
 +  * vex2difx to create an input file setup to be used with vlbi_fake ​
 +  * <​del>​Complete Support for VDIF on module directories</​del>​
 +  * Refactoring the mpifxcorr/​mk5daemon libraries and merging functionality
 +  * Move to a binary format for pcal files and name by datastream not by antenna name
 +  * vex2 support (relies on vex2 finalization and some support from field systems)
 +  * On-the-fly application of calibration (bandpasses etc) with appropriate accountability
 +  * Modernize sniffer
 +  * cleanup of DiFX error messages
 +  * K5 format support (CP)
 +  * Space VLBI
 +    * Baseline dependent averaging
 +  * eVLBI gui
 +  * Generic startup scripts (CR)
 +    * Documentation of startup philosophy
 +  * Shifted FT (moving channel boundary to edge of band) (AD)
 +  * DiFX2fits uv shifting in trunk (JM/AD)
 +  * Auditing / refactoring the amplitude scaling
 +  * Pulsar predictor support in "​polyco"​
 +  * Get rid of printing to screen in mark5access (change to be called and printed from elsewhere)
 +  * Get Eric to improve AIPS PCAL (multi tone)
 +  * Improve the runtime predictor (make better use of the new cluster configuration file, better estimate true load)
 +  * Restitching contiguous zoom bands into wider bands (e.g. for ALMA)
difx/difx-todolist-2012.txt · Last modified: 2015/10/21 10:08 (external edit)