Black failure on Travis-ci

[erwanp]

🐛 Describe the bug

error: cannot format /home/travis/build/radis/radis/radis/levels/partfunc_cdsd.py: INTERNAL ERROR: Black produced code that is not equivalent to the source.  Please report a bug on https://github.com/psf/black/issues.  This diff might be helpful: /tmp/blk_xcyo54jw.log

Example : https://travis-ci.com/github/radis/radis/jobs/379490900#L266

But works perfectly with Local Black run (Python 3.6, Python 3.8)

💡 Possible solutions

No idea at this point, but we may have to run Travis locally :

https://stackoverflow.com/questions/21053657/how-to-run-travis-ci-locally

[erwanp]

Actually I could reproduce the bug using the Latest Black version (20.18)

psf/black#1658

--- src
+++ dst
@@ -1299,11 +1299,11 @@
       body=
         Expr(
           value=
             Constant(
               value=
-                'Calculate Partition Function from energy levels (and maybe export\na tabulated database).\n\nwarning::\nZPE (zero point energy) must be the same in the Line Database and the\nEnergy levels database. See the\n:ref:`Configuration file <label_lbl_config_file>`.\n\nParameters\n----------\n\nenergy_levels: filename\npath to energy levels (to calculate Partition Function) for ``isotope``\n\nisotope: int\nwhich isotope we\'re dealing with. Default ``1``. In the current implementation\nonly isotope 1 and 2 are defined.\n\nlevelsfmt: ``\'cdsd-p\'``, ``\'cdsd-pc\'``, ``\'cdsd-pcN\'``, ``\'cdsd-hamil\'``, or ``None``\nthe format of the Energy Database, and in particular how ``Evib`` and ``Erot``\nhave been calculated. A vibrational level in the CDSD (p,c,J,N) nomenclature\ncan be defined for levels that share a same (p), (p,c) or (p,c,N), where\n``p`` is the polyad number, ``c`` is the Wang symmetry number, and ``N``\nis the ranking index of a (p,c,J) group. Default ``\'cdsd-pc\'``.\n\nIf ``None``, dont label the levels. Wont be able to use the EnergyDatabase to fetch\nvibrational energies for lines, however it can still be used to\ncalculate Partition functions independently from a Spectrum calculation\n\nOther Parameters\n----------------\n\nuse_cached: ``True``, ``False``, or ``\'regen\'``, ``\'force\'``\nif ``True``, use (and generate if doesnt exist) a ``.h5`` file.\nIf ``\'regen\'``, regenerate cache file. If ``\'force\'``, raise an error\nif file doesnt exist. Default ``True``\n\nuse_json: boolean\ndeprecated. Better use h5 now.\n\nNotes\n-----\n\nDatabase format:\n\nTaskhkun database updated with ranking number (n) & total rank (N) of\nblock, Evib and Erot (cm-1)  and jref\n\nFor nonequilibrium, different strategies exist so as how to assign rotational and vibrational\nenergies in a CDSD database. See the E. Pannier "Limits on CO2 Nonequilibrium\nmodel" article for a discussion on that.\n\nExample of table format::\n\n# calculated rovibrational energy levels of 12C16O2\n# =================================================\n# S.Tashkun, Zuev Institute of Atmospheric Optics, Tomsk, Russia\n# date: 17.03.2017\n#\n# zero point energy ZPE (cm-1) =  2531.828\n#\n# p = 2v1 + v2 + 3v3 - polyad number\n# j - rotational quantum number\n# c - Wang symmetry (1-\'e\'; 2-\'f\')\n# N - ranking number of energy levels of (p,j,c) blocks\n# n - total rank of (p,j,c) blocks\n#\n# Calculation limitations:\n# pmax = 40\n# jmax = 300\n# Ecut = 44600 cm-1\n# ---------------\np\tc\tj\tN\tn\tE\tEvib\tErot\tjref\n0\t1\t0\t1\t1\t0.000\t0.000\t0.000\t0\n0\t1\t2\t1\t1\t2.341\t0.000\t2.341\t0\n0\t1\t4\t1\t1\t7.804\t0.000\t7.804\t0\n0\t1\t6\t1\t1\t16.389\t0.000\t16.389\t0\n0\t1\t8\t1\t1\t28.095\t0.000\t28.095\t0\n\nSee an example in `test/files/co2_cdsd_hamiltonian_fragment.levels <https://github.com/radis/radis/blob/develop/radis/test/files/cdsd_hitemp_09_fragment.txt>`\n\nSee Also\n--------\n\n:py:class:`~radis.levels.partfunc_cdsd.PartFuncCO2_CDSDtab`',  # str
+                'Calculate Partition Function from energy levels (and maybe export\na tabulated database).\n\nwarning::\nZPE (zero point energy) must be the same in the Line Database and the\nEnergy levels database. See the\n:ref:`Configuration file <label_lbl_config_file>`.\n\nParameters\n----------\n\nenergy_levels: filename\npath to energy levels (to calculate Partition Function) for ``isotope``\n\nisotope: int\nwhich isotope we\'re dealing with. Default ``1``. In the current implementation\nonly isotope 1 and 2 are defined.\n\nlevelsfmt: ``\'cdsd-p\'``, ``\'cdsd-pc\'``, ``\'cdsd-pcN\'``, ``\'cdsd-hamil\'``, or ``None``\nthe format of the Energy Database, and in particular how ``Evib`` and ``Erot``\nhave been calculated. A vibrational level in the CDSD (p,c,J,N) nomenclature\ncan be defined for levels that share a same (p), (p,c) or (p,c,N), where\n``p`` is the polyad number, ``c`` is the Wang symmetry number, and ``N``\nis the ranking index of a (p,c,J) group. Default ``\'cdsd-pc\'``.\n\nIf ``None``, dont label the levels. Wont be able to use the EnergyDatabase to fetch\nvibrational energies for lines, however it can still be used to\ncalculate Partition functions independently from a Spectrum calculation\n\nOther Parameters\n----------------\n\nuse_cached: ``True``, ``False``, or ``\'regen\'``, ``\'force\'``\nif ``True``, use (and generate if doesnt exist) a ``.h5`` file.\nIf ``\'regen\'``, regenerate cache file. If ``\'force\'``, raise an error\nif file doesnt exist. Default ``True``\n\nuse_json: boolean\ndeprecated. Better use h5 now.\n\nNotes\n-----\n\nDatabase format:\n\nTaskhkun database updated with ranking number (n) & total rank (N) of\nblock, Evib and Erot (cm-1)  and jref\n\nFor nonequilibrium, different strategies exist so as how to assign rotational and vibrational\nenergies in a CDSD database. See the E. Pannier "Limits on CO2 Nonequilibrium\nmodel" article for a discussion on that.\n\nExample of table format::\n\n# calculated rovibrational energy levels of 12C16O2\n# =================================================\n# S.Tashkun, Zuev Institute of Atmospheric Optics, Tomsk, Russia\n# date: 17.03.2017\n#\n# zero point energy ZPE (cm-1) =  2531.828\n#\n# p = 2v1 + v2 + 3v3 - polyad number\n# j - rotational quantum number\n# c - Wang symmetry (1-\'e\'; 2-\'f\')\n# N - ranking number of energy levels of (p,j,c) blocks\n# n - total rank of (p,j,c) blocks\n#\n# Calculation limitations:\n# pmax = 40\n# jmax = 300\n# Ecut = 44600 cm-1\n# ---------------\np       c       j       N       n       E       Evib    Erot    jref\n0       1       0       1       1       0.000   0.000   0.000   0\n0       1       2       1       1       2.341   0.000   2.341   0\n0       1       4       1       1       7.804   0.000   7.804   0\n0       1       6       1       1       16.389  0.000   16.389  0\n0       1       8       1       1       28.095  0.000   28.095  0\n\nSee an example in `test/files/co2_cdsd_hamiltonian_fragment.levels <https://github.com/radis/radis/blob/develop/radis/test/files/cdsd_hitemp_09_fragment.txt>`\n\nSee Also\n--------\n\n:py:class:`~radis.levels.partfunc_cdsd.PartFuncCO2_CDSDtab`',  # str
             )  # /Constant
         )  # /Expr
         FunctionDef(
           args=
             arguments(

[erwanp]

Error was due to this change in Black : psf/black#1288

trailing commas are handled diffrently in 20.18 and 19.10. Different black version were used on the server and locally. Finally fixed in 7fcc0a7