Empty subplot causes ghostscript to hang

[adigitoleo]

Julia 1.7.3
GMT (system install) 6.3
GMT.jl 0.42.4 (also on master)

I have found that this doesn't affect Windows.

Repro:

using GMT
subplot(grid = (3, 3), dims = (10, 10))
gmtend()

also happens for:

using GMT
gmtbegin("test.pdf")
subplot(grid = (3, 3), dims = (10, 10))
gmtend()

After this ghostscript needs to be killed manually (kill -9).

Neither of these GMT scripts seem to cause this, they throw an error as expected:

#!/bin/sh
gmt begin test pdf
  gmt subplot begin 3x3 -Baf -BWSen -F10/10
gmt end

#!/bin/sh
gmt begin test pdf
  gmt subplot begin 3x3 -Baf -BWSen -F10/10
  gmt subplot end
gmt end

I'm not able to follow how the subplot context is handled, I guess it happens through one of the global variables. Anyway, just found this while playing around with subplot for some examples.

[adigitoleo]

OK I see now that the context is managed through IamSubplot:

julia> gmtbegin("test.pdf")

julia> subplot(grid = (3, 3), dims = (10, 10))

julia> GMT.IamSubplot[1]
true

Maybe gmtend is not checking for this.

[adigitoleo]

Hmm, there is a try/catch, not sure why it isn't stopping this:

https://github.com/GenericMappingTools/GMT.jl/blob/master/src/common_options.jl#L3554-L3556

[joa-quim]

Can reproduce on Windows too. But here's what I think is happening. In plain GMT to use subplot one has to do:

gmt begin
     subplot begin
     subplot end
gmt end

but in GMT.jl when using subplot it implicitly sets the gmtbegin() ... gmtend() steps. So a gmtend() should not be used (as in your example) because it ends the ps without having finished the subplot(...).
For some reason ghostscript doesn't like the produced ps falls in some infinite loop.

Will see how to catch this situation.

[joa-quim]

Should be fixed by #939

[adigitoleo]

Yep, this is solved for me on master now, throws an subplot [ERROR]: No subplot information file! which is expected.