Correct y limits in ProfileViewerState when min same as max

[pllim]

This fixes a display limit problem we see over at Cubeviz when cube is all ones.

import numpy as np
from astropy import units as u
from specutils import Spectrum1D

from jdaviz import Cubeviz

sp = Spectrum1D(flux=np.ones((7, 8, 9)) * u.nJy)

cubeviz_helper = Cubeviz()
cubeviz_helper.load_data(sp, data_label='test_cube')
cubeviz_helper.show()

Without this patch:

With this patch:

🐱

[dhomeier]

Sorry for the late reply. A good choice for a default range still is somewhat arbitrary, since I cannot conceive any real data where this case would apply. But I've tested this branch with some other examples, and setting a range of at least a few percents would probably be good to have it visible in the tick marks.
I think it should be possible to extend test_limits to cover this case?

[pllim]

@dhomeier good point about very small values. I updated the code. How does it look now?

[pllim]

Where is test_limits? I cannot find it.

[dhomeier]

I had just edited the suggestion to a slightly different version to account for negative values as well, which is probably now hidden in the resolved.

[dhomeier]

glue/glue/viewers/profile/tests/test_state.py

Line 130 in 7481caf

def test_limits(self):

[pllim]

@dhomeier , can u please re-suggest the neg fix so commit will credit you properly? Also feel free to push directly to this branch if that is easier. Thanks!

[pllim]

I don't think the Windows job failures are related?

[dhomeier]

Suggestion is easier for now, but if you don't find time to write a test, let me know, and I can probably look into that later.

[dhomeier]

No, the image loader failures I have seen elsewhere.

[pllim]

Added a couple of tests. Please re-review. Thanks!

[dhomeier]

Thanks! I'd parametrise that test, but cannot push to your branch.

I have pushed to my own fork, so could open a PR against this, phew.

[pllim]

Hmm I enabled push by maintainer, so weird it won't let you push. Did you work off my branch? If you rebase, then that might mess up the history, and not let you push without a force.

[dhomeier]

No, I had reset to e9df453, so should be on the same history, and it is not failing on that afaict:

To github.com:pllim/glue.git
! [remote rejected] pllim-fix-yminmax-same -> pllim-fix-yminmax-same (permission denied)

[pllim]

PR to my branch works too. Pls ping me there. Thx

[dhomeier]

pllim#1

[dhomeier]

I created the feature branch from glue-viz/glue main, not sure if that matters.

[pllim]

@dhomeier , I merged your PR to my PR, so your changes is in this PR now. Thanks!

[dhomeier]

Thanks, LGTM now!

[pllim]

Thanks for your help!

[astrofrog]

Thanks!