numba `TypingError` when running large-tokamak regression test

[jonmaddock]

When trying to run the large-tokamak regression test, I get a TypingError from numba:

**************************************************************************************************************
 
   Equality constraints : 26
 Inequality constraints : 00
      Total constraints : 26
    Iteration variables : 44
         Max iterations : 200
       Figure of merit  : +01  -- minimise major radius.
  Convergence parameter : 1.00E-07
 
 **************************************************************************************************************
Traceback (most recent call last):
  File "/home/jon/anaconda3/envs/process/bin/process", line 33, in <module>
    sys.exit(load_entry_point('process', 'console_scripts', 'process')())
  File "/home/jon/code/process/process/main.py", line 650, in main
    Process(args)
  File "/home/jon/code/process/process/main.py", line 127, in __init__
    self.run_mode()
  File "/home/jon/code/process/process/main.py", line 204, in run_mode
    self.run.run()
  File "/home/jon/code/process/process/main.py", line 363, in run
    self.run_scan(self.solver)
  File "/home/jon/code/process/process/main.py", line 459, in run_scan
    self.scan = Scan(self.models, solver)
  File "/home/jon/code/process/process/scan.py", line 22, in __init__
    self.run_scan()
  File "/home/jon/code/process/process/scan.py", line 36, in run_scan
    ifail = self.doopt()
  File "/home/jon/code/process/process/scan.py", line 56, in doopt
    ifail = self.optimiser.run()
  File "/home/jon/code/process/process/optimiser.py", line 47, in run
    evaluators = Evaluators(self.models, x)
  File "/home/jon/code/process/process/evaluators.py", line 28, in __init__
    self.caller = Caller(models, x)
  File "/home/jon/code/process/process/caller.py", line 20, in __init__
    self.call_models(x)
  File "/home/jon/code/process/process/caller.py", line 65, in call_models
    self.models.physics.physics()
  File "/home/jon/code/process/process/physics.py", line 998, in physics
    current_drive_variables.cboot * self.bootstrap_fraction_sauter()
  File "/home/jon/code/process/process/physics.py", line 4627, in bootstrap_fraction_sauter
    dcsa(
  File "/home/jon/.local/lib/python3.10/site-packages/numba/core/dispatcher.py", line 468, in _compile_for_args
    error_rewrite(e, 'typing')
  File "/home/jon/.local/lib/python3.10/site-packages/numba/core/dispatcher.py", line 409, in error_rewrite
    raise e.with_traceback(None)
numba.core.errors.TypingError: Failed in nopython mode pipeline (step: nopython frontend)
Internal error at <numba.core.typeinfer.CallConstraint object at 0x7f9b8ff168f0>.
Failed in nopython mode pipeline (step: nopython rewrites)
Constant inference not possible for: $222build_string.6 + $const220.5

File "../../../../process/physics.py", line 776:
def tpf(j, triang, sqeps, fit=1):
    <source elided>

    raise RuntimeError(f"fit={fit} is not valid. Must be 1, 2, or 3")
    ^

During: resolving callee type: type(CPUDispatcher(<function tpf at 0x7f9b909ddd80>))
During: typing of call at /home/jon/code/process/process/physics.py (511)

Enable logging at debug level for details.

File "../../../../process/physics.py", line 511:
def dcsa(j, nr, rmajor, bt, triang, ne, ni, tempe, tempi, mu, rho, zef, sqeps):
    <source elided>
    zz = zef[j - 1]
    zft = tpf(j, triang, sqeps)
    ^

Obviously this needs to be fixed (possible as a result of #3088), but also how did this get through the CI?

My numba version is 0.56.4.

[timothy-nunn]

Working on numba 0.57+

In 0.57 the following warning occurs:

/Users/timothynunn/process/process/physics.py:522: NumbaExperimentalFeatureWarning: Use of isinstance() detected. This is an experimental feature.