Make function parenPrecedence greater than unary negative, postive

[Alex-Jordan]

This may be controversial, so consider this PR as up for discussion.

In the forums, there is this thread: https://webwork.maa.org/moodle/mod/forum/discuss.php?d=4790

Prior to this commit, Formula("-sqrt(x)") will render as -\left(\sqrt{x}\right) because the function application has lower precedence than negation. The commit changes that, so that this example renders as -\sqrt{x}. And similarly for any -function(x).

I can't think of a reason to keep it the way it is. The change seems to me to be an aesthetic improvement. And also a pedagogical improvement, since occasionally we get students wondering why WW is printing those parentheses, and asking if they are needed. Of course, I may be missing something.

[drgrice1]

Is this something that we want to target for ww-2.16? I think it is a small change, and is a nice aesthetic improvement.

Note that there is another instance in which parentheses are rendered where they are not needed as well. That is the case that parserAssignment.pl is used. Things like x=-3 render as x=(-3). It would be nice to not have the parentheses there also.

[taniwallach]

The change works for me, and I think it is a good improvement.
I think this can be merged for WW 2.16.
@dvpc - Any comments on potential adverse effects of the change to precedence.

[taniwallach]

I labeled as Do Not Merge Yet as I think any feedback from @dpvc about potential adverse consequences would be helpful before we do the merge. If he does not have time to look/comment - maybe someone else has some in-depth understanding of the potential implications of the precedence change.

[taniwallach]

@drdrew42 @drgrice1 - what do you think about this?

[drgrice1]

I think this is a good change. @dpvc suggested it to @Alex-Jordan, although it would be good to get his official review. I am not sure how in depth he had thought it out when he suggested it.

[pstaabp]

The develop branch needs to be merged in again to the PR. I'm not able to view anything because of the discrepancy between MathJax 2.7.X (which is in this PR) and 3.X which is develop.

[drgrice1]

@pstaabp: You can merge the develop branch into the pull request branch after checking it out. It is not necessary to merge the develop branch into this pull request. That should become part of your workflow when testing pull requests.

[pstaabp]

@drgrice1 good call. I was trying figure out how I could test with Alex doing this.

[pstaabp]

Overall, looks good.

Is this something that we want to target for ww-2.16? I think it is a small change, and is a nice aesthetic improvement.

Note that there is another instance in which parentheses are rendered where they are not needed as well. That is the case that parserAssignment.pl is used. Things like x=-3 render as x=(-3). It would be nice to not have the parentheses there also.

I would agree with adding this as well, but we can do that at a later time.

I noted that parserImplicitEquation.pl on Formula("x=-3") returns x=-3 as expected.

[pstaabp]

I approve.

[dpvc]

I don't think this should cause any problems.

[dpvc]

Note that there is another instance in which parentheses are rendered where they are not needed as well. That is the case that parserAssignment.pl is used. Things like x=-3 render as x=(-3). It would be nice to not have the parentheses there also.

This is unrelated to the function application precedence, and can be controlled by the showExtraParens flag:

Context()->flags->set(showExtraParens=>0);

but this will also remove other extra parentheses that make operator-operand connections painfully clear.

I've also made a pull request that will prevent these particular parentheses when parserAssignment.pl is used.

[drgrice1]

I have tried setting the showExtraParens flag to 0 before (and just tested it again), and still get x=(-3). Your pull request does remove the additional parentheses though.

[dpvc]

This test

loadMacros("parserAssignment.pl");
Context("Numeric")->flags->set(showExtraParens => 0); 
parser::Assignment->Allow;
TEXT(Formula("x = -3")->string);

produces x = -3 for me (without the PR), while with showExtraParens => 1 I get x = (-3).

[drgrice1]

Yes, but if the student enters x=-3 as the answer it is displayed as x=(-3). For example using:

loadMacros("PGstandard.pl", "MathObjects.pl", "PGML.pl", "parserAssignment.pl");

Context("Numeric")->flags->set(showExtraParens => 0);
parser::Assignment->Allow;

$f1 = Formula("x = -3");

BEGIN_PGML
Enter [`[$f1]`]: [______]{$f1}
END_PGML

[dpvc]

Sorry, the answer checker sets showExtraParens internally (it resets several flags during processing of student answers), so that is the reason.

[drgrice1]

Yeah, I see that now. Well, your pull request works for student answers also.