ocamlpro / scilint Goto Github PK

-->[1,2,
-->3,4]
ans =

1.    2.  
3.    4.  

-->[1,2,3,4]
ans =

1.    2.    3.    4.  

Scilint should warn for "," at the end of line.

disp should never be used with several arguments

Among the following examples, only the first one is valid, scilab should warn about the others.
[x, y] = (x, y)
[x, y] = [x, y]
[x, y] = x, y

I read somewhere on the internet that ~expr was the same as ~~~~expr, which is not, it is a repeated application of the ~ operator, and it happens that applying a boolean negation an odd (and > 1) number of times is useless since it can be rewritten as a single one. Applying ~ twice is ok, it is a cast to bool (~~3 is T), but applying it four or more times is also useless.

Scilint should warn about a number of nested applications of ~ strictly superior to 2.
Scilint should also probably warn about a number of applications of - strictly superior to 1.

It would be nice if scilint could export results to the firehose format:
https://github.com/fedora-static-analysis/firehose

2+3*%i

String "box-1", line 1, characters 2-6:
Error: multiplication does not work between values of types real and complex
Hint: you can try redefining %s_m_s

lycee()
exp(10)/factorielle(10)

String "box-2", line 1, characters 8-23:
String "factorielle.sci", line 16, characters 4-16:
Error: primitive factorial does not work on a value of type real
Hint: you can try redefining %s_factorial
⇊Insert a new box after this one.

I found a probably erroneous piece of code:
clear X,Y ;
which is equivalent to:
clear X
Y
and not at all what the programmer probably meant:
clear ('X','Y')

Scilint should warn about this however there is no simple way from the AST.
Three options seem possible to me:

• warn in a more general way about any sequence of two instructions located on the same line, this should detect this one but may be too restrictive
• detect this directly in the parser (probably easier inside my upcoming experimental parser)
• implement some sort of simple textual matching on the source

indexes usually start at 1