2011年5月24日Scala Language Specification在第6.12.3节中发现了一个拼写错误here。这是邮件列表中的acknowledged。
中缀运营商的实际优先级是什么?
答案 0 :(得分:24)
我认为通过测试来计算它会很有趣,我编写了以下代码在REPL中执行。给出:
val ops = List(
"letter", "|", "^",
"&", "<", ">", "!",
"+", "-", "*", "/", "%", "?",
"=?", // add ? to prevent assignment
":?" // add ? to prevent right-association
)
首先生成一个使用和测试运算符的中间scala文件。
import java.io._
// generate a class with all ops operators defined
// where operator combines in a way we can figure out the precedence
val methods = ops.map("""def %s(o: Op) = Op("["+o.v+v+"]")""".format(_))
val body = methods.mkString("\n")
val out = new PrintWriter(new FileWriter("Op.scala"))
out.println("case class Op(v: String) {\n%s\n}".format(body))
// generate tests for all combinations and store in comps
// Op(".") op1 Op(".") op2 Op(".") v returns "[[..].]" when op2 > op1
// returns "[.[..]]" when op1 <= op2
def test(op1: String, op2:String) = {
"""("%s","%s") -> (Op(".") %s Op(".") %s Op(".")).v.startsWith("[[")""".
format(op1, op2, op1, op2)
}
val tests = for (op1 <- ops; op2 <- ops) yield { test(op1, op2) }
out.println("val comps = Map[(String, String), Boolean](%s)".format(
tests.mkString(",\n")))
out.close
然后加载Op类,运行测试并加载comps
:load Op.scala
// order operators based on tests
val order = ops.sortWith((x,y) => comps(x -> y))
// if op1 or op2 don't have higher precedence, they have the same precedence
def samePrecedence(op1: String, op2: String) =
!comps(op1 -> op2) && !comps(op2 -> op1)
def printPrecedenceGroups(list: List[String]): Unit = {
if (list != Nil) {
val (same, rest) = list.span(op => samePrecedence(op, list.head))
println(same.mkString(" "))
printPrecedenceGroups(rest)
}
}
printPrecedenceGroups(order)
打印:
letter
|
^
&
! =?
< >
:?
+ -
* / %
?
因此需要使用< >
切换与规范的主要区别{。{1}}。