Re: FIFO
On Monday, 13 May 2024 at 15:07:39 UTC, Andy Valencia wrote: On Sunday, 12 May 2024 at 22:03:21 UTC, Ferhat Kurtulmuş wrote: https://dlang.org/phobos/std_container_slist.html This is a stack, isn't it? LIFO? Ahh yes. Then use dlist Thank you. I read its source, and was curious so I wrote a small performance measurement: put 10,000 things in a FIFO, pull them back out, and loop around that 10,000 times. My FIFO resulted in: real0m1.589s user0m1.585s sys 0m0.004s And the dlist based one: real0m4.731s user0m5.211s sys 0m0.308s Representing the FIFO as a linked list clearly has its cost, but I found the increased system time interesting. OS memory allocations maybe? The code is spaghetti, fifo/dlist, but it seemed the easiest way to see the two API's being used side by side: version(fifo) { import tiny.fifo : FIFO; } else { import std.container.dlist : DList; } const uint ITERS = 10_000; const uint DEPTH = 10_000; void main() { version(fifo) { auto d = FIFO!uint(); } else { auto d = DList!uint(); } foreach(_; 0 .. ITERS) { foreach(x; 0 .. DEPTH) { version(fifo) { d.add(x); } else { d.insertBack(x); } } foreach(x; 0 .. DEPTH) { version(fifo) { assert(x == d.next()); } else { assert(x == d.front()); d.removeFront(); } } } } thank you for sharing the results. Everything I read about queues recommends doublylinked lists. With your array based implementatio if you are consuming the elements faster than pushing new elements, your array buffer never resize which is costly. This should explain why your array based queue is more performant.
Re: FIFO
On Monday, 13 May 2024 at 15:07:39 UTC, Andy Valencia wrote: Representing the FIFO as a linked list clearly has its cost, but I found the increased system time interesting. OS memory allocations maybe? I know you want FIFO, I usually keep this on hand for fixed size LIFO; It can easily convert to FIFO and doesn't use LinkedList: ```d class LIFO(T) { T * element; size_t length, size; this(size_t s) { element = cast(T*)new T[s]; length = s; } auto rewind() => size = length; bool empty() => !size; auto front() => element[size - 1]; auto popFront() => --size; auto pop() => empty ? T(0) : element[--size]; alias push = insertFront; auto insertFront(T value) in(size < length, "Stack is overflow!") => element[size++] = value; auto ref opDollar() => length; auto ref opIndex(size_t i) => element[i]; auto ref opSlice(size_t first, size_t last) => element[first..last]; } unittest { enum test = 7; auto stack = new LIFO!size_t(test); assert(!stack.size); foreach (prime; 1..test + 1) { stack.insertFront(prime); } assert(stack.size == test); // == 7 stack.writeln(": ", stack.length); // [7, 6, 5, 4, 3, 2, 1]: 7 stack[$/2].writeln("-", stack[0]); // 4-1 stack.rewind(); stack.size.write(": ["); // 10: while (auto next = stack.pop) { if (next == 1) next.writeln("]"); else next.write(", "); } stack.size.writeln; // 0 stack.rewind(); assert(stack.size == test); } SDB@79
Re: FIFO
On Sunday, 12 May 2024 at 22:03:21 UTC, Ferhat Kurtulmuş wrote: https://dlang.org/phobos/std_container_slist.html This is a stack, isn't it? LIFO? Ahh yes. Then use dlist Thank you. I read its source, and was curious so I wrote a small performance measurement: put 10,000 things in a FIFO, pull them back out, and loop around that 10,000 times. My FIFO resulted in: real0m1.589s user0m1.585s sys 0m0.004s And the dlist based one: real0m4.731s user0m5.211s sys 0m0.308s Representing the FIFO as a linked list clearly has its cost, but I found the increased system time interesting. OS memory allocations maybe? The code is spaghetti, fifo/dlist, but it seemed the easiest way to see the two API's being used side by side: version(fifo) { import tiny.fifo : FIFO; } else { import std.container.dlist : DList; } const uint ITERS = 10_000; const uint DEPTH = 10_000; void main() { version(fifo) { auto d = FIFO!uint(); } else { auto d = DList!uint(); } foreach(_; 0 .. ITERS) { foreach(x; 0 .. DEPTH) { version(fifo) { d.add(x); } else { d.insertBack(x); } } foreach(x; 0 .. DEPTH) { version(fifo) { assert(x == d.next()); } else { assert(x == d.front()); d.removeFront(); } } } }
Re: What prevents ImportC from using .h directly?
On Sunday, 12 May 2024 at 21:34:30 UTC, Chris Piker wrote: So, why does ImportC need *.c files exclusively? I'm sure it solves a problem, but I don't know what that problem is. Support for headers has been worked on, but had to be reverted: https://issues.dlang.org/show_bug.cgi?id=23479
Re: Find homography in D?
On Sunday, 21 April 2024 at 14:57:33 UTC, Paolo Invernizzi wrote: Hi, Someone can point me to a D implementation of the classical OpenCV find homography matrix? Thank you, Paolo Now, we can do image stitching using DCV. It needs improvements though. https://github.com/libmir/dcv/tree/master/examples/imagestitchinghomography
Re: FIFO
On Saturday, 11 May 2024 at 23:44:28 UTC, Andy Valencia wrote: I need a FIFO for a work scheduler, and nothing suitable jumped out at me. I wrote the following, but as a newbie, would be happy to receive any suggestions or observations. TIA! [...] I don't know your use case, maybe you have a similar problem I had to solve years ago. https://forum.dlang.org/post/xktftztisodpngcow...@forum.dlang.org