20 May visualization meeting

Data Management Camera-DM Visualization
#1

Agenda:

Notes:

  • Discussion of @tony_johnson’s diagram:

  • To speed processing, each input node in the DAQ cluster will perform image corrections (gains, etc.).

  • To avoid storage bandwidth (to save money), the visualization ingester will make an in-memory FITS file.

  • I&T data storage: one FITSfile per raft (9 CCDs).

  • @tony_johnson has posted a docfile and a demo of the image provider (called checkImage, I think) here:

  • https://github.com/lsst-camera-visualization/visualization-server

  • For now, it just delivers fake images.

  • How to combine extension headers into the primary header.

  • If all we care about is geometry (good enough for now), we can assume a regular pattern of sizes and gaps. It may be that we just need four configuration files, one for each CCD vendor (ITL and E2V), with and without overscans.

  • Seth Digel has a keyword doc that @tony_johnson will post it.

  • What to do in La Serena will be a discussion topic at the June LSST meeting. We need to show our visualization plan.

  • Suggestion (not being worked on now): Instead of FITS, pass FireFly its own internal image representation.

Action: @ywang: Look at Stuart’s Load Image JS command.

Action: @Everyone: In the absence of a meeting (it’s summertime), post progress and status to Community and/or GitHub.

#2

Also would be good to discuss:

The “Get new image” functionality (more precisely find out if new image exists and if so where), or at least a first version of it, is now in github. There is a README describing it here:

Stuart’s progress on running FireFly at SLAC

#3

I checked the recent code that Stuart committed and I found no function called load image. Which branch did he commit his code?

3 June visualization meeting
#4

I did not commit it yet. Sorry about the delay. I will post or commit something today.

#5

I made a new branch “slm-cleanup” containing my modified files. cmd.js has the load_image() command. Also I used eslint and tidy to clean things up a bit. We should settle on a scheme for both the html and js formatting. It seems eslint is a reasonable choice but I’d like to hear what others would use. Also note the server end configuration (as a tomcat application) has to be configured to load files.