@taro Taro is correct. Vote up. That should be

f:= unapply(%, x, y);

There is no Maple syntax where a comma is used as something other than as a separator, and it's never just empty space that's being separated.

Yeah, those pages should be proofread. I'd guess that the program that generates those web pages treats % as a special character, and whoever typed up the page forgot that.

@tomleslie Vote Up. Thank you for noticing the odd-function thing.  The sign of the argument has something to do with it. Also, I think that there may be some difference between polynomials that the user enters (such as b-a+c) and those that are automatically generated such as those that appear as the arguments to ln and arctan in the Question. I'm just guessing about that last sentence.

@nm The critical thing that makes the difference between your examples is the lexicographic ordering of x0 and the other variable. I have no doubt about that. The exact nature of that difference is harder to classify: I'm not saying that every polynomial will appear in lexicographic order. The sign of the polynomial also has something to do with it, and the sign depends somewhat on the lexicographic order. I expanded my Answer, and I hope that that clarifies the matter somewhat.

@zack94 If you post any Maple-related Question, I'll give you 5 reputation.

@Joe Riel Hmm, I'm aware that the value that my program above returns is roughly proportional to kernelopts(stacklimit). It seemed unlikely to me that that knowledge would help the OP, so I didn't mention it. 

So, now that you've mentioned it, do you know of any aspects of the procedure itself (such as its size) that would change the count?

@rlopez Why do you think that she wants a numeric solution?

@Preben Alsholm Yes, that consistency is a good reason. I was disappointed by the arrival of this kernelopts option because I thought that it should apply to all constants. Nonetheless, it is more of a nuisance to me to have to use evalf every time that I have a result with Pi that should be floating point.

It is a time-wasting nuisance that I see your Questions both here and on StackOverflow. I wish that you wouldn't simultaneously ask the same Questions in both places. Perhaps you could wait a reasonable time, and if you don't get a suitable Answer in one place, then ask in the other?

@Preben Alsholm Why do you prefer floatPi= false?

Yes, those bugs exists for me also---Maple 2016 x64 Win 8.1.

@acer The following is the first sentence of the fifth paragraph of Description at ?Digits. I think that this is poorly worded, and may cause confusion between the roles of interface(displayprecision) and Digits.

The interface(displayprecision) command controls the number of decimal places to be displayed, which is determined by Digits.

@Axel Vogt 

Axel, what is your Digits set to for the above computation?

@awass &where is mostly just a symbol. You can put & in front of any symbol, and it becomes a "neutral operator" (see ?&). There's a tiny amount of code behind &where, which you can see by showstat(`&where`). The only purpose of &where is to compactly format returned solutions in terms of variable substitutions. It often occurs in results returned by pdsolve. Essentially the same thing can be done with Eval (with a capital E!).

@MortenZdk

What documentation are you referring to?

@san There will always be some size of matrix at which your monitor is incapable of displaying it "in stander matrix form," as you say. Isn't that obvious??? Either buy a bigger monitor, set the Zoom factor lower, or be content with looking at your matrix one piece at a time.

You can set the Zoom factor from the View menu.

If A is a matrix, you can see rows 15 - 20, columns 10 - 15 of A by

A[15..20, 10..15];