Re: [Swig-user] Python, fscanf and graphical toolkit conflict

[David F. <dav...@gm...>]

> The problem is likely not in in fscanf, but rather in the calls to printf.
>  I suggest re-writing foo_read_values to return the values that it reads,
> rather than printf-ing them.  Then, check the values that are returned to
> python, and see if they are correct.

So I've made the test: the returned value is incorrect. I also printf
the expected value. The format is different, but it prints correct
value.
I also compare the read-value with the expected, it confirm the value is wrong.

I attach the modified files, and here are the outputs:

In pure C:
>From C: v0=1.000000, expected=1.000000, n1=1
>From C: v1=2.000000, expected=2.000000, n2=1
>From C: v2=3.000000, expected=3.000000, n3=1
>From C: Read-float has expected value (with a 0.001 tolerance).
>From C: Read-float has expected value (with a 0.001 tolerance).
>From C: Read-float has expected value (with a 0.001 tolerance).

In C/Python, not importing matplotlib:
>From C: v0=1.000000, expected=1.000000, n1=1
>From C: v1=2.000000, expected=2.000000, n2=1
>From C: v2=3.000000, expected=3.000000, n3=1
>From C: Read-float has expected value (with a 0.001 tolerance).
>From C: Read-float has expected value (with a 0.001 tolerance).
>From C: Read-float has expected value (with a 0.001 tolerance).
>From Python: returned value is: 3.000000, expected: 3.000000

In C/Python, importing matplotlib:
>From C: v0=1,000000, expected=1,000000, n1=1
>From C: v1=-<a random value here>, expected=2,000000, n2=0
>From C: v2=0,000000, expected=3,000000, n3=0
>From C: Read-float has expected value (with a 0.001 tolerance).
>From C: Read-float does not have expected value (with a 0.001 tolerance).
>From C: Read-float does not have expected value (with a 0.001 tolerance).
>From Python: returned value is: 0.000000, expected: 3.000000

Best,
David