digitalmars.D - Built-in arrays as output ranges
- Steve Teale <steve.teale britseyeview.com> Feb 07 2010
- Andrei Alexandrescu <SeeWebsiteForEmail erdani.org> Feb 07 2010
- Steve Teale <steve.teale britseyeview.com> Feb 07 2010
Not a bug, but would it be reasonable to expand a little on the documentation
for an output range, perhaps along the lines:
r.put(e) puts e in the range (in a range-dependent manner) and advances to the
popFront position in the range. Successive calls to r.put add elements to the
range. put may throw to signal failure.
When using a built-in array as an output range, to do anything useful it is
necessary to take a reference to the original array before using put. The
original array will be nibbled away by put operations (see std.array). For
example:
int[] a = [ 1,2,3 ];
int[] b = a;
a.put(-a[0]);
a.put(-a[0]);
writefln("([%s] [%s]", b, a);
// [ -1, -2, 3] [ 3 ]
Feb 07 2010
Steve Teale wrote:Not a bug, but would it be reasonable to expand a little on the documentation for an output range, perhaps along the lines: r.put(e) puts e in the range (in a range-dependent manner) and advances to the popFront position in the range. Successive calls to r.put add elements to the range. put may throw to signal failure. When using a built-in array as an output range, to do anything useful it is necessary to take a reference to the original array before using put. The original array will be nibbled away by put operations (see std.array). For example: int[] a = [ 1,2,3 ]; int[] b = a; a.put(-a[0]); a.put(-a[0]); writefln("([%s] [%s]", b, a); // [ -1, -2, 3] [ 3 ]
Right. Speaking of which, I think it's sensible to always leave the test for empty range in, even in release unsafe builds. void put(T, E)(ref T[] range, E element) if (!isSomeString!(T[])) { enforce(!range.empty, "Attempting to put in an empty array"); *range.ptr = element; range.popFront(); } enforce() will never be disabled. As an aside, I just realized I haven't implemented put for strings yet, and also that I'd promised a check in this weekend. Andrei
Feb 07 2010
enforce() will never be disabled. As an aside, I just realized I haven't implemented put for strings yet, and also that I'd promised a check in this weekend. Andrei
Actually, thinking about this overnight, I'm a bit unhappy about giving the impression that a built-in array can serve as an output range. It really isn't true unless you never want to see the output again. If you do, some data structure is required, either a loose combination of an array and an unprotected reference to its original state (arrays a and b), or something more explicit like: struct arrayOutputRange(T) { T[] array; uint pos; this(uint sx) { ... } void put(T val) { ... } } Steve
Feb 07 2010
Steve Teale wrote:enforce() will never be disabled. As an aside, I just realized I haven't implemented put for strings yet, and also that I'd promised a check in this weekend. Andrei
Actually, thinking about this overnight, I'm a bit unhappy about giving the impression that a built-in array can serve as an output range. It really isn't true unless you never want to see the output again. If you do, some data structure is required, either a loose combination of an array and an unprotected reference to its original state (arrays a and b), or something more explicit like: struct arrayOutputRange(T) { T[] array; uint pos; this(uint sx) { ... } void put(T val) { ... } } Steve
copy(source, target) does make sense for arrays as output ranges. Since target is passed by value, your copy will see what's been copied. Andrei
Feb 08 2010








Andrei Alexandrescu <SeeWebsiteForEmail erdani.org>