So far I have marked the traced center positions in DS9, but I couldn't find a function in display.ds9 to overplot a rectangular area in a DS9 image. I do have a function that reconstructs an image from the extracted spectrum and spatial profile what should fulfill the same purpose as overlaying a FiberTraceSet on a flat field...

Created an ipython notebook in drp_stella/examples/ (tickets/PIPE2D-25). Data are in drp_stella_data data/PFS/postISRCCD/2016-01-12/v0000005/PFFAr2.fits (tickets/PIPE2D-25)

import lsst.afw.display as afwDisplay

disp = afwDisplay.Display(1)

disp.mtv(...)
disp.line([(x0, y0), (x0, y0+10), (x0 + 20, y0 + 10), (x0 + 20, y0), (x0, y0)])

(note: this is the way to use ds9; please don't import lsst.afw.display.ds9 directly)

The dataID in the notebook is incorrect:

RuntimeError: No unique lookup for ['taiObs', 'dateObs', 'field', 'spectrograph', 'arm'] from {'filter': 'r', 'ccd': 5, 'category': 'A', 'visit': 5, 'site': 'S'}: 0 matches

Using site=F works.

Fixed the site error, sorry about that...

The notebook appears to work, but I have some concerns that need to be addressed:

1. For the loop marking the centres of the trace there appears to be an off-by-one error in x (the + signs all appear to the right of the trace). The plotting code appears to be correct, so is this a problem in your fiberTrace extraction? Please check.
2. Why do you set point?
3. Why is this not encapsulated in a function?
4. Why did you choose to mark the centre with a set of crosses rather than drawing the ridgeline?
5. You need to use

   with display.Buffering():

   whenever using display writes.

For markFiberTraceInMask:

1. Did this have to be written in C++?
2. The value of the masked image plane is hard coded to 1
3. The code that sets the mask calls getArray; there is no need to go through the numpy interface
4. The object passed in is a MaskedImage, not a Mask but only the Mask is modified
5. The inner loop does the entire traversal

   maskedImage->getMask()->getArray()[ y + iY ][ x ] = 1;

   that should be outside the loop

In the notebook,

1. what is

   display0.setMaskPlaneColor("CROSSTALK", "orange")

   supposed to do? There is no such mask plane in the data, and it'd be unrelated to fibre traces if there was.

2. Why do you say

   afwDisplay.setDefaultBackend("ds9" if True else "virtualDevice")

   Isn't that the default it display_ds9 is setup?

Also (and out of scope for this ticket but worrying)

1. The routine calcMinCenMax() appears very inefficient, making copies of many ndarrays. Also, why is it a free function not part of a FiberTrace (given that when used you have to unpack the FiberTrace)
2. The indentation in Math.cc is incorrect
3. The lines in FiberTraces.cc exceed the permitted length of 110 characters

Merged into master