diff --git a/README.md b/README.md
index 5b84e79..3bbdecf 100644
--- a/README.md
+++ b/README.md
@@ -63,6 +63,14 @@ In stage 2 each buffer is populated by the HEAD of the coresponding file
   - [ ] push operation requires heapify function
 - [ ] buffer i/o operations
 
+# Quirks
+
+the reader, writer, buffer, etc is a mess, could be much simpler, clearer, more readable, and less error prone if not for the "reading in chunks of bytes"
+
+It creates a need for two separate buffers for reading, three if counting internal buffers of stream operations in c++. Unncessairly increases complexity, and so on.
+
+I'm not really sure if i'm even doing it correctly and as intended, cause it's so messy
+
 # Analysis
 
 #todo
diff --git a/makefile b/makefile
index 4545a71..5ff5a85 100644
--- a/makefile
+++ b/makefile
@@ -18,7 +18,10 @@ build/run/1.in:
 run: build/main build/run/1.in
 	build/main -i build/run/1.in -d build/run/tmp
 
+run_alt: build/main build/run/1.in
+	build/main -i build/run/1.in -d build/run/tmp -n 11 -b 100
+
 debug: build/main tests/1.in
 	lldb -- build/main
 
-.PHONY: clean run debug test generate
+.PHONY: clean run debug test generate run_alt
diff --git a/sort.cpp b/sort.cpp
index 8e78ce2..045b16a 100644
--- a/sort.cpp
+++ b/sort.cpp
@@ -3,7 +3,11 @@
 #include "config.hpp"
 #include "reader.hpp"
 #include "writer.hpp"
+#include <exception>
 #include <fstream>
+#include <queue>
+#include <stdexcept>
+#include <string>
 
 // returns amount of records read
 size_t read_chunk(Reader& reader, SubBuffer& buff)
@@ -30,13 +34,15 @@ void write_to_file(const std::string& filename, SubBuffer& buff, size_t n)
     }
 }
 
-void create_initial_runs(const std::string& input_filename, const std::string& directory, Buffer& main_buffer)
+size_t create_initial_runs(const std::string& input_filename, const std::string& directory, Buffer& main_buffer)
 {
     auto buff = main_buffer.divide(1)[0];
 
     std::ifstream in_stream(input_filename);
     Reader input_reader(in_stream);
-    for (size_t run_index = 0;; run_index++) {
+    size_t run_index;
+
+    for (run_index = 0;; run_index++) {
         size_t size = read_chunk(input_reader, buff);
         if (size == 0) {
             break;
@@ -44,13 +50,141 @@ void create_initial_runs(const std::string& input_filename, const std::string& d
         std::sort(buff.begin(), buff.begin() + size);
         write_to_file(directory + "/" + std::to_string(run_index) + ".run", buff, size);
     }
+
+    return run_index;
+}
+
+void write_block(Writer& w, SubBuffer& buff, size_t n)
+{
+    for (size_t i = 0; i < n; i++) {
+        w.write(buff[i]);
+    }
+}
+
+void k_way_merge_and_write(std::vector<Reader>& input_readers, Writer& output_writer, std::vector<SubBuffer>& buffers)
+{
+    // for debug purposes
+    std::cout << "buffers size: " << buffers.size() << " input_readers: " << input_readers.size() << '\n';
+    for (size_t i = 0; i < input_readers.size(); i++) {
+        std::cout << "reader " << i << ": " << input_readers[i].read().value() << '\n';
+    }
+
+    struct HeapElement {
+        Record record;
+        size_t source_index;
+        // priority queue is a max heap by default, so comparison is backwards
+        bool operator<(const HeapElement& other) const
+        {
+            return !(record < other.record);
+        }
+    };
+    using MinHeap = std::priority_queue<HeapElement>;
+    MinHeap pq;
+
+    struct BufferState {
+        size_t items_total = 0;
+        size_t current_idx = 0;
+    };
+
+    std::vector<BufferState> buffer_states(input_readers.size());
+
+    // initial fill
+    for (size_t i = 0; i < input_readers.size(); i++) {
+        SubBuffer& input_buffer = buffers[i];
+        buffer_states[i].items_total = read_chunk(input_readers[i], input_buffer);
+        buffer_states[i].current_idx = 0;
+
+        if (buffer_states[i].items_total == 0) {
+            continue;
+        }
+
+        pq.push({ input_buffer[0], i });
+        buffer_states[i].current_idx++;
+    }
+
+    size_t output_index = 0;
+    SubBuffer& output_buffer = buffers[buffers.size() - 1];
+
+    while (!pq.empty()) {
+        HeapElement next = pq.top();
+        pq.pop();
+
+        output_buffer[output_index] = next.record;
+        output_index++;
+
+        if (output_index == output_buffer.size()) {
+            write_block(output_writer, output_buffer, output_index);
+            output_index = 0;
+        }
+
+        size_t source_index = next.source_index;
+
+        if (buffer_states[source_index].current_idx >= buffer_states[source_index].items_total) {
+            buffer_states[source_index].items_total = read_chunk(input_readers[source_index], buffers[source_index]);
+            buffer_states[source_index].current_idx = 0;
+            if (buffer_states[source_index].items_total == 0) {
+                continue;
+            }
+        }
+
+        pq.push({ buffers[source_index][buffer_states[source_index].current_idx], source_index });
+        buffer_states[source_index].current_idx++;
+    }
+}
+
+// I think more or less works correctly now
+// creates input readers, and output writer for a merge to be performed by another function
+// if not every file can be processed at once, splits them
+size_t run_merge_pass(std::vector<SubBuffer>& buffers, std::string current_dir, std::string next_dir, size_t file_count)
+{
+    const size_t K = buffers.size() - 1; // Merge factor (e.g., 3)
+
+    size_t new_runs = 0;
+    for (size_t i = 0; i < file_count; i += K) {
+        // std::cout << "pass 0 run " << i << '\n';
+        // create input readers, to turn into a function i would have to some complicated ownership semantics, and i don't want to bother with that
+        std::vector<std::ifstream> in_streams;
+        std::vector<Reader> input_readers;
+        in_streams.reserve(K);
+        input_readers.reserve(K);
+        for (size_t j = 0; j < K; j++) {
+            std::string filename = current_dir + "/" + std::to_string(i + j) + ".run";
+            in_streams.emplace_back(std::ifstream(filename));
+            input_readers.emplace_back(Reader(in_streams[j]));
+        }
+
+        // create the output writer -||-
+        std::string filename = next_dir + "/" + std::to_string(new_runs) + ".run";
+        std::filesystem::path dirPath = std::filesystem::path(filename).parent_path();
+        std::filesystem::create_directories(dirPath);
+        std::ofstream out_stream(filename);
+        Writer output_writer(out_stream);
+
+        k_way_merge_and_write(input_readers, output_writer, buffers);
+        new_runs++;
+    }
+
+    return new_runs;
 }
 
 int sort_file(const Configuration& opts)
 {
     Buffer main_buffer(opts.n * opts.b);
+    size_t file_count;
 
-    create_initial_runs(opts.input_file, opts.directory, main_buffer);
+    file_count = create_initial_runs(opts.input_file, opts.directory + "/pass0", main_buffer);
 
+    auto buffers = main_buffer.divide(opts.n);
+    for (int i = 0;; i++) {
+        std::string current_dir = opts.directory + "/pass" + std::to_string(i);
+        std::string next_dir = opts.directory + "/pass" + std::to_string(i + 1);
+        file_count = run_merge_pass(buffers, current_dir, next_dir, file_count);
+        if (file_count == 1) {
+            // #todo move the resulting file to the output dir
+            break;
+        }
+    }
+
+    // #todo cleanup tmp files
     return 0;
 }