coordinate dependencies list isn't always in the correct order

[mfiaudrin-epsyl]

Hi @jwasikpsnc,
I have seen a bug using the application. It occurs rarely, that's why we didn't seen it before.

Let's reproduce it:

• uri : imas:hdf5?path=/work/imas/shared/imasdb/ITER/3/135011/7/
• Add a plot using path : #core_sources/source[0]/profiles_1d[0]/total_ion_energy
• Switch to 1D plot mode
• Upating time cursor is crashing the app

After having investigate it, the problem occurs because coordinate dependencies are not listed in the expected order in this example.

I show you why in the coordinate list:

• rho_tor_norm [15, 986, 50]
• time [986]
• source [15]

I receive from BE rho_tor_norm.coordinates = ['time', 'source'] whereas it should be ['source', 'time'] because the shape of rho_to_norm is not [986, 15, 50] but [15, 986, 50].

Can you please ensure the BE provide the correct order in all cases ?

[mfiaudrin-epsyl]

Hi @jwasikpsnc,
After some other investigations, it looks that BE is always returning the coordinate dependencies list in the inverse order of what I expected. If you confirm that it's the case, I can handle it in FE side

[jwasikpsnc]

Hi @mfiaudrin-epsyl
I have just returned from long sick-leave.
What happened here?
Is the issue still present?
I didn't checked what's going on with coordinates order, should i do that?
Was this issue present on develop, or other branches?

[mfiaudrin-epsyl]

Hi @mfiaudrin-epsyl I have just returned from long sick-leave. What happened here? Is the issue still present? I didn't checked what's going on with coordinates order, should i do that? Was this issue present on develop, or other branches?

Hi @jwasikpsnc I hope you've recovered,
The problem has been resolved. I realized that I hadn't handled the list of coordinates dependencies correctly. They was in the reverse order that I thought they was.
At first, I thought their order was random but it seems they are always in the same order. So I managed them in FE in the order BE sends them, which fixed the bug I had noticed.
The problem was in develop since we have "coordinate[i].coordinate", but I only discovered it last week, because in most of our examples, we had at most a single coordinate dependency.