maxima parser not smart enough

[vdelecroix]

As found in ​this post on ask.sagemath.org the function desolve_system can fail with weird error message

sage: t = var('t')
sage: f = function('f')(t)
sage: g = function('g')(t)
sage: de1 = diff(f,t) == f*sin(t)
sage: de2 = diff(g,t) == 0
sage: desolve_system([de1, de2], [f,g])
Traceback (most recent call last):
<ipython-input-1-58fe9994eeb3> in <module>()
      2 de1 = diff(f,t) == f*sin(t);
      3 de2 = diff(g,t) == Integer(0);
----> 4 desolve_system([de1, de2], [f,g])

.../sage/calculus/desolvers.pyc in desolve_system(des, vars, ics, ivar)
    838     for i, sol in enumerate(soln):
    839         print(sol)
--> 840         soln[i] = sol.sage()
    841     if ics is not None:
    842         ivar_ic = ics[0]
...
ValueError: The name "+" is not a valid Python identifier.

The reason is that the .sage() method is not able to convert the maxima's answer back into Sage

f(_SAGE_VAR_t) = ilt((%i laplace(f(_SAGE_VAR_t), _SAGE_VAR_t, g2579 + %i)
 - %i laplace(f(_SAGE_VAR_t), _SAGE_VAR_t, g2579 - %i) + 2 f(0))/(2 g2579), 
g2579, _SAGE_VAR_t)

[egourgoulhon]

The bug seems to be still there, see this recent ​ask.sagemath question.

[chapoton]

simpler way to trigger the same error:

        sage: from sage.calculus.calculus import dummy_laplace
        sage: s,t = var('s,t')
        sage: f = function('f')
        sage: dummy_laplace(f(t),t,s+1)

[chapoton]

not totally convinced by the "fix"

---

New commits:

​fd3436b

third args in laplace and inverse_laplace

[vdelecroix]

What about a doctest?

[git]

Branch pushed to git repo; I updated commit sha1. New commits:

​759a65d	Merge branch 'u/chapoton/23445' in 9.6.rc0
​c34e155	add doctest

[chapoton]

voila

mais il reste un probleme de non-oubli (permanence) de la fonction g

[mkoeppe]

author name

[chapoton]

as said in comment:8, there remains a problem about not forgetting g, which causes a failing doctest