Kernel dies if an additional subset map is added

[michaelgrund]

Hey guys,

I recently tried to generate a figure in a jupyter notebook which consists of two maps (A and B). For that I first generated a map (A) using fig.coast(...) which works well. For the second map (B) I first use shift_origin to move it a little bit right. Then I use again fig.coast(...) to generate map B. At this point the kernel dies and it's not possible to continue. Any ideas what happens here?

The issue is similar to #514, however, not exactly the same.

import pygmt

fig = pygmt.Figure()

lon = 19
lat = 64

fig.coast(region = 'g',
          projection = 'E' + str(lon) + '/' + str(lat) + '/5c',
          resolution = 'c', 
          shorelines = '1/thinnest,black', 
          frame = ['30g0'],
          area_thresh = ['5000'],
          G = 'lightgray',
          C = 'lightgray')

fig.shift_origin(xshift = '3c')

fig.coast(region = 'g',
          projection = 'E' + str(lon) + '/' + str(lat) + '/5c',
          resolution = 'c', 
          shorelines = '1/thinnest,black', 
          frame = ['30g0'],
          area_thresh = ['5000'],
          G = 'lightgray',
          C = 'lightgray')

[welcome]

👋 Thanks for opening your first issue here! Please make sure you filled out the template with as much detail as possible. You might also want to take a look at our contributing guidelines and code of conduct.

[MarkWieczorek]

I think that shift_origin() only works with gmt 6.1.1. This is supposed to be released in a few days, but if you are impatient, you can install from source. On macOS, this is easy to do with brew install gmt --HEAD.

[michaelgrund]

I think that shift_origin() only works with gmt 6.1.1. This is supposed to be released in a few days, but if you are impatient, you can install from source. On macOS, this is easy to do with brew install gmt --HEAD.

Hmm, I created several other maps in the same manner, however, using different projections and everything worked so far without having 6.1.1

[MarkWieczorek]

I confirm that this crashes the kernel on my system using the most up to date versions of gmt and pygmt (as of about a week ago...)

coast [WARNING]: Fill/clip continent option (-G) may not work for this projection.
coast [WARNING]: If the antipode (0/-5.52561e+09) is in the ocean then chances are good it will work.
coast [WARNING]: Otherwise, avoid projection center coordinates that are exact multiples of 199 degrees.
[1]    75191 segmentation fault  ipython

And, the crash is unrelated to shift_origin. If you simply call fig.coast twice, the kernel dies....

[michaelgrund]

And, the crash is unrelated to shift_origin. If you simply call fig.coast twice, the kernel dies....

That's strange since in another figure I perform nearly the same, except that I use two different projections for the subsets (https://github.com/michaelgrund/GMT-plotting/blob/master/009_paper_GR2020/pygmt_jn_fig_2/GR_2020_Fig_2.ipynb). It seems that it crashes if multiple times the same projection is used...

[seisman]

I can confirm that it's an upstream GMT bug.

@PaulWessel Please see if you can fix it in GMT.

[PaulWessel]

I take it the equivalent shell script works fine?

[seisman]

I take it the equivalent shell script works fine?

Yes, please also note that the two fig.coast() calls are exactly the same, but the first one works and it crashes at the 2nd one.

[PaulWessel]

Also sinks MATLAB.

[seisman]

@michaelgrund The bug was fixed in the upstream GMT and a new GMT release (GMT 6.1.1) will be available in the next few days.

We will close the issue when PyGMT bumps the minimum required GMT version to 6.1.1.

[michaelgrund]

Great to hear!

[weiji14]

Closing as we've bumped up the minimum GMT version to 6.1.1 in #577. The code should work now to produce this plot:

[michaelgrund]

yes after updating to 6.1.1 it works, thanks for your efforts!!!