Bug 3868 – It would be nice to have a function which read a file lazily using a range

It would be great if there were a function which returned a range to a file, reading it in as you moved through the range. Presumably, it would do some buffering, but it wouldn't be a case of reading in the whole file at a time and hopefully would be more friendly and powerful than reading in a single byte, or int, or whatever like you'd do with read() with streams. It also avoids having to explicitly read in a portion of the file at a time like you'd do with readBlock(). In addition, it would make it possible to use various of the standard algorithms directly on files. It might also be useful to write to a file with a range, but I'm not acquainted well enough with output ranges to know how well it would work, or if it would be as useful as reading with ranges. It would probably be useful though. In any case, I really like the idea of being to read a file with a range. Maybe there's a reason that I'm not aware of which would make it a really bad idea, but I think that it would be a good addition to phobos.

Shouldn't std.stdio.File.byChunk() do this?

byChunk uses opApply. We need to transform it into a range; using opApply alone severely limits the applicability of byChunk. Here's a nice potential application of byChunk: import std.stdio; void main(string[] a) { enforce(a.length == 3); auto f1 = File(a[1]), f2 = File(a[2]); immutable bufsize = 1024 * 1024; return equal(f1.byChunk(bufsize), f2.byChunk(bufsize)); } One other idea suggested by the above is to implement the algorithm found in diff in std.algorithm. Then we can write a diff program in a dozen line of code, using general components.

byChunk is now a proper range. Is everyone happy that this is resolved?

Yah. Thanks for revisiting this!