Apply user_functions to vecFcode too

[alkino]

When functions inside solve block are not inlined sympy cannot find them. The problem is already fixed for vecJcode, do the same.

Related to #927

[bbpbuildbot]

Logfiles from GitLab pipeline #75392 (:white_check_mark:) have been uploaded here!

Status and direct links:

• ✅ spack_setup
• ✅ build:intel
• ✅ build:nvhpc
• ✅ test:intel
• ✅ test:nvhpc

[cattabiani]

may you add a CI test that crashes with the previous version? It should be simple: with sympy 1.9 beta(x) was transformed to beta(x,x). This probably cannot compile.

[alkino]

This patch is not fixing this bug. But the bug that not inlined functions are not found.

[cattabiani]

ok, it was another bug. I sincerly do not know how this impacts nmodl. I do not know what it should do. I think you need someone else for the review

[alkino]

Without the patch, the following file fail

NEURON {
    SUFFIX glia
}

STATE {
    C1
    C2
}

BREAKPOINT {
    SOLVE kstates METHOD sparse
}

FUNCTION alfa(v(mV)) {
    alfa = exp(v)
}

FUNCTION tau(v(mV)) {
    tau = exp(-v)
}

KINETIC kstates {
    ~ C1 <-> C2 (alfa(v), tau(v))
}

Because functions alfa and tau are not known, so sympy return

vecFcode=['F[0] = /* Not supported in C: */
/* alfa */
/* tau */
-X[0]*dt*alfa(v) - X[0] + X[1]*dt*pow(tgamma(v), 2)/tgamma(2*v) + old_C1',
'F[1] = /* Not supported in C: */
/* alfa */
/* tau */
X[0]*dt*alfa(v) - X[1]*dt*pow(tgamma(v), 2)/tgamma(2*v) - X[1] + old_C2']

And our internal parser don't know C comments

[cattabiani]

it looks like it actually solved the problem. Are you sure to the left you do not have F[0] and F[1] and somewhere someone leaves the '? Atm I forgot how C1 <-> C2() translates to a pde.

[pramodk]

it looks like it actually solved the problem.

That's my understanding as well. By providing user_functions, sympy no longer produce C comments and avoids errors in NMODL parser.

@alkino : about the failing mod file we had - maybe we can add that as a unit test? (in the unit test, we will not run Inline visitor and then it should cover the use case)

[bbpbuildbot]

Logfiles from GitLab pipeline #76033 (:white_check_mark:) have been uploaded here!

Status and direct links:

• ✅ spack_setup
• ✅ build:intel
• ✅ build:nvhpc
• ✅ test:intel
• ✅ test:nvhpc

[pramodk]

we can merge #937 to this branch first?

[bbpbuildbot]

Logfiles from GitLab pipeline #76067 (:no_entry:) have been uploaded here!

Status and direct links:

• ✅ spack_setup
• ✅ build:intel
• ✅ build:nvhpc
• ✅ test:intel
• ✅ test:nvhpc

[bbpbuildbot]

Logfiles from GitLab pipeline #76073 (:white_check_mark:) have been uploaded here!

Status and direct links:

• ✅ spack_setup
• ✅ build:intel
• ✅ build:nvhpc
• ✅ test:intel
• ✅ test:nvhpc

[iomaganaris]

LGTM