digitalmars.D.bugs - [Issue 11811] New: Add "eval" to phobos
- d-bugmail puremagic.com (46/46) Dec 24 2013 https://d.puremagic.com/issues/show_bug.cgi?id=11811
https://d.puremagic.com/issues/show_bug.cgi?id=11811 Summary: Add "eval" to phobos Product: D Version: D2 Platform: All OS/Version: All Status: NEW Severity: enhancement Priority: P2 Component: Phobos AssignedTo: nobody puremagic.com ReportedBy: monarchdodra gmail.com The function documented here: http://dlang.org/function.html#interpretation Or as suggested here: http://forum.dlang.org/thread/wqbbequembgvchmwmhnt forum.dlang.org#post-l7n550:2415ku:241:40digitalmars.com The idea is to simply *force* the CTFE evaluation of a variable, without declaring an enum. This can be useful when a *variable* needs to be initialized by a function, and said function has a certain cost, but all it's arguments are known at compile time. for example: int[] inc = eval!(iota(0, 10).array()); or string s = eval!(toRoman(1998)); or int big = eval!(fib(20)); This could have been an alternative for the implementation of octal, for example. It can also be used to pass as an argument to a function. For example: void foo(T)() pure nothrow { assert(someConditon, eval!(format("error %s because of %s", T.stringof, T.sizeof)); } Here, as you can see, the error string is evaluated at compile, so there is no overhead during runtime. It does not require an extra variable to store the enum, and is useable in a pure and nothrow context. This makes initialization both more efficient, and easier: All in a single line, and no overhead. I'm not sure where to put this: Either typecons or conv, I'd say. Ideally, it would be in its own package, as this is a purely convenience function that requires nothing. -- Configure issuemail: https://d.puremagic.com/issues/userprefs.cgi?tab=email ------- You are receiving this mail because: -------
Dec 24 2013