Possible error in FEQParse

[M-M-1980]

I think there's a bug in the FEQParse (Fortran Equation Parser) program. Using the example 'scalar_with_scalar_eval.f90', I get different results depending on the order of the components:

y = x-sqrt(x)+100

   x                    y

0.000000 -100.0000

4.000000 -98.00000

y = -sqrt(x)+100+x

   x                    y

0.000000 100.0000

4.000000 102.0000

I am using the latest available version of FEQParse.

[fluidnumericsJoe]

Hey there @M-M-1980 , I've been able to reproduce the issue and have added a unit test to start work on this. Hoping to have this sorted out in the coming days

[fluidnumericsJoe]

Hey @M-M-1980 , sorry for the long delay here. A lot of spinning plates and moving parts these days. Looks like we've got a PR for resolving this issue open at #46

[M-M-1980]

Hello fluidnumerics-joe.
Thank you very much for the changes. Initial tests are producing correct results, but I still have a request for error-proofing when editing patterns. Example on the same pattern (repeated minus sign '-'):
Y=x - - sqrt(x)+100
and
Y= - - sqrt(x)+100+x
The results are astonishing.
Best regards

[fluidnumericsJoe]

What value is produced by writing an infix equation in such an odd way? What's stopping you from using parentheses?

[M-M-1980]

My request concerned cases when the program user accidentally makes this type of mistake (for example, enters the minus sign twice). I will try to protect myself against such cases in my software by pre-analyzing the entered pattern.

[fluidnumericsJoe]

Such an equation is not even valid fortran. I'll consider putting in a sanitizer that catches invalid infix notation, but will not support processing sloppy input