Range Warnings begin<end

[MichaelPFrey]

based on http://forum.openscad.org/Infamous-warnings-td28054.html

• remove the now redundant code in value.cc
• consider generalizing warning
  • (just focusing on literals for now)
  • Warn for
    • for(i=[10:1]) -> swap, deprecated
    • for(i=[1:-1:10]) -> warning
    • for(i=[1:0:10]) -> bad range warning (not very specific, but "fair enough" with the line number)
  • correct (do not warn)
    • for(i=[0:10]) -> fine
    • for(i=[10:-1:1]) -> fine
  • not addressed with this PR
    • expression containing variables like for(i=[0:a]) for(i=[a:0]) for(i=[0:a:10]) for(i=[a:1:10]) for(i=[0:1:a])
    • non-numbers like
      • for(r=[1:true:5]) echo(r);
      • for(r=[1:"test":5]) echo(r);
      • for(r=["a":"z"]) echo(r);
• test cases
  • (check list from above)
  • manually review changed test cases
• move the warning to a NOINLINE function to minimize memory footprint?
• rephrase warnings for consistency

[nophead]

Isn't for(i=[10:-1:1]) the legal way to iterate backwards? I thought what was deprecated was automatically changing the implicit increment to -1 when the range was backwards. Not banning backwards ranges altogether.

[MichaelPFrey]

concate-tests has something interesting:

echo(concat([3, 4, true, 6], [4:1:3]));

for-tests has something interesting:

// Illegal step value
for(r=[1:true:5]) translate([r*10-60,50,0]) cylinder(r=r);

some more examples:

[MichaelPFrey]

Isn't for(i=[10:-1:1]) the legal way to iterate backwards?

Yes it is - I should separate out "should not create warnings" from "should create warnings" and "not my concern right now" in my to do list to be more clear.