added incomplete feature _FS_SPILL_BUFFER

include/fsvec.hpp

@@ -15,7 +15,7 @@ struct file_back_buffer
 	file_back_buffer();
 	file_back_buffer(std::size_t cap);
 	file_back_buffer(const file_back_buffer& c) = delete;
-	inline file_back_buffer(file_back_buffer&& m) : inner(std::move(m.inner)){}
+	file_back_buffer(file_back_buffer&& m);
 
 	void push_buf(byte* buf, std::size_t len);
 	bool back(byte* buf, std::size_t len) const;
@@ -46,7 +46,7 @@ template<typename T>
 struct file_vector
 {
 	inline file_vector() : file_vector(file_back_buffer::DEFAULT_CAP){}
-	inline file_vector(std::size_t cap) : inserter(file_back_buffer(cap)), len(0), current_back(std::vector(sizeof(T))) {current_back.resize(sizeof(T));}
+	inline file_vector(std::size_t cap) : inserter(file_back_buffer(cap)), len(0), current_back(std::vector<unsigned char>(sizeof(T))) {current_back.resize(sizeof(T));}
 	inline file_vector(const file_vector<T>& c) = delete;
 	inline file_vector(file_vector<T>&& m) : inserter(std::move(m.inserter)), len(m.len), current_back(std::move(m.current_back)){}
 
@@ -74,8 +74,10 @@ struct file_vector
 		if(!inserter.pop_n(sizeof(T))) panic("pop_back(): 0 elements");
 		len-=1;
 	}
+
+	inline const std::size_t size() const { return len; }
 private:
-	mutable std::vector<unsigned char> current_back; // what an awful hack...
 	file_back_buffer inserter;
-	std::uint64_t len;
+	std::size_t len;
+	mutable std::vector<unsigned char> current_back; // what an awful hack...
 };

include/shuffle.hpp

@@ -5,6 +5,10 @@
 #include <algorithm>
 #include <iostream>
 
+#include <fsvec.hpp>
+
+#include <shuffle3.h>
+
 namespace rng {
 	template<typename T, typename R>
 	inline void shuffle(R& rng, span<T> span)
@@ -23,7 +27,12 @@ namespace rng {
 	inline void unshuffle(R& rng, span<T> span)
 	{
 		if(!span.size()) return;
+
+#ifdef _FS_SPILL_BUFFER
+		file_vector<std::size_t> rng_values(span.size());
+#else
 		std::vector<std::size_t> rng_values(span.size());
+#endif
 
 		std::cout << " -> unshuffling " << span.size() << " objects...";
 		for(std::size_t i=span.size()-1;i>0;i--)

include/shuffle3.h

@@ -14,6 +14,10 @@ extern "C" {
 #define _FORCE_INLINE __attribute__((gnu_inline)) extern inline
 #endif
 
+//** Features **//
+
+//#define _FS_SPILL_BUFFER /* Use a file-backed buffer when unshuffling in cases of too high memory usage. Will cause massive slowdowns but can stop OOMs when unshuffling large files */
+
 extern const char* _prog_name;
 
 #ifdef __cplusplus

src/fsvec.c

@@ -4,7 +4,7 @@
 
 int fvec_new(fvec_t* restrict obj, const char* path)
 {
-	obj->backing = fopen(path, "wb");
+	obj->backing = fopen(path, "w+b");
 	if(!obj->backing) {
 		perror("fvec_new(): Failed to open file");
 		return 0;
@@ -60,7 +60,7 @@ int fvec_get_whole_buffer(const fvec_t* restrict obj, void* _buffer, size_t _sz)
 	while ( w<_sz && (c=fread(buffer+w, 1, sz-w, obj->backing))>0 ) w+=c;
 	if (w!=_sz) {
 		perror("Corrupted buffer state, aborting");
-		panic("Cannot continue");
+		panic("Cannot continue on FD %d (%p)", fileno(obj->backing), obj->backing);
 	}
 
 	return 1;

src/fsvec.cpp

@@ -3,6 +3,7 @@
 #include <fsvec.hpp>
 #include <uuid.hpp>
 #include <fsvec.h>
+#include <debug.h>
 
 #define FB file_back_buffer
 using std::size_t;
@@ -13,17 +14,25 @@ struct temp_file
 {
 	inline temp_file(const temp_file& c) = delete;
 
-	inline temp_file(temp_file&& m) : name(std::move(m.name)) {}
-	inline temp_file() : name(uuid::generate().to_string()){}
+	inline temp_file(temp_file&& m) : name(std::move(m.name)) { m._full_path.clear(); }
+	inline temp_file() : name(uuid::generate().to_string()+"-s3"){}
 	inline temp_file(const char* name) : name(name) {}
 	inline temp_file(std::string&& name) : name(name) {}
 
-	inline ~temp_file() =  default;
+	inline ~temp_file()
+	{
+		if(!_full_path.empty() && fs::exists(_full_path) ) {
+			D_dprintf("tfile removing: %s", _full_path.c_str());
+			fs::remove(_full_path);
+		}
+	}
 	
 	inline const fs::path& full_path() const
 	{
-		if(_full_path.empty())
-			_full_path = fs::canonical( fs::temp_directory_path() / name );
+		if(_full_path.empty()) {
+			_full_path = fs::absolute( fs::temp_directory_path() / name );
+			D_dprintf("tfile path: %s", _full_path.c_str());
+		}
 		return _full_path;
 	}
 	inline const std::string& base_name() const { return name; }
@@ -40,11 +49,6 @@ struct FB::impl
 	temp_file file;
 
 	fvec_t backing;
-
-	inline ~impl() 
-	{
-		fvec_close(&backing);
-	}
 };
 
 FB::FB(size_t cap) : inner(std::make_unique<FB::impl>())
@@ -56,6 +60,10 @@ FB::FB(size_t cap) : inner(std::make_unique<FB::impl>())
 
 }
 FB::FB() : FB(DEFAULT_CAP){}
+FB::~FB()
+{
+	fvec_close(&inner->backing);
+}
 
 void FB::push_buf(byte* buf, size_t len)
 {
@@ -71,3 +79,27 @@ bool FB::pop_n(size_t len)
 {
 	return !!fvec_pop_end(&inner->backing, len);
 }
+
+extern "C" void _fb_run_tests()
+{
+	file_vector<int> test;
+	int r0,r1=0;
+	for(int i=0;i<10;i++) {
+		D_dprintf("push: %d", (10-i));
+		test.push_back(10-i);
+		r1+= (10-i);
+	}
+	D_dprintf("r1: %d", r1);
+	r0=0;
+	while(test.size())
+	{
+		r0+=test.back();
+		D_dprintf("back: %d", test.back());
+		test.pop_back();
+	}
+	D_dprintf("r0: %d", r0);
+
+	if(r0!=r1) panic("fb failed test: %d, %d", r0, r1);
+
+	D_dprintf("test successful");
+}

src/main.c

@@ -19,6 +19,8 @@ _Static_assert(sizeof(float)==sizeof(uint32_t), "float is not 32 bits");
 
 const char* _prog_name;
 
+extern void _fb_run_tests();
+
 noreturn help_then_exit()
 {
 	fprintf(stderr, "Try passing `-h`\n");
@@ -41,6 +43,7 @@ int main(int argc, char** argv)
 {
 	_prog_name = argv[0];
 
+	//_fb_run_tests();
 	work_args_t parsed;
 
 	if( !argv[1] || *(argv[1]) != '-') help_then_exit();