diff --git a/src/core/index.rs b/src/core/index.rs
index 54699a3fcf..cd2f84dd4c 100644
--- a/src/core/index.rs
+++ b/src/core/index.rs
@@ -43,7 +43,7 @@ impl fmt::Debug for Index {
    }
 }
 
-type DirectoryPtr = Box<Directory>;
+pub type DirectoryPtr = Box<Directory>;
 
 #[derive(Clone)]
 pub struct Index {
@@ -127,7 +127,7 @@ impl Index {
     pub fn publish_segment(&mut self, segment: &Segment) -> Result<()> {
         {
             let mut meta_write = self.metas.write().unwrap();
-            meta_write.segments.push(segment.segment_id.clone());
+            meta_write.segments.push(segment.segment_id);
         }
         self.save_metas()
     }
@@ -148,7 +148,6 @@ impl Index {
             }
             meta_write.segments.push(merged_segment.id());
         }
-        // TODO use logs
         self.save_metas()
     }
 
@@ -201,10 +200,11 @@ impl Index {
     pub fn save_metas(&mut self,) -> Result<()> {
         let mut w = Vec::new();
         {
-            let metas_lock = self.metas.read().unwrap() ;
+            let metas_lock = self.metas.read().unwrap();
             try!(write!(&mut w, "{}\n", json::as_pretty_json(&*metas_lock)));
         };
-        try!(self.rw_directory())
+        try!(self
+            .rw_directory())
             .atomic_write(&META_FILEPATH, &w[..])
     }
 }
diff --git a/src/core/segment_reader.rs b/src/core/segment_reader.rs
index a403419f2f..3fcbea0653 100644
--- a/src/core/segment_reader.rs
+++ b/src/core/segment_reader.rs
@@ -107,7 +107,7 @@ impl SegmentReader {
         
         let positions_data = segment
                 .open_read(SegmentComponent::POSITIONS)
-                .unwrap_or(ReadOnlySource::Anonymous(Vec::new()));
+                .unwrap_or(ReadOnlySource::empty());
         
         let schema = segment.schema();
         Ok(SegmentReader {
diff --git a/src/datastruct/fstmap.rs b/src/datastruct/fstmap.rs
index 8c77419ab5..6aead53823 100644
--- a/src/datastruct/fstmap.rs
+++ b/src/datastruct/fstmap.rs
@@ -61,7 +61,7 @@ pub struct FstMap<V: BinarySerializable> {
 
 fn open_fst_index(source: ReadOnlySource) -> io::Result<fst::Map> {
     Ok(fst::Map::from(match source {
-        ReadOnlySource::Anonymous(data) => try!(Fst::from_bytes(data).map_err(convert_fst_error)),
+        ReadOnlySource::Anonymous(data) => try!(Fst::from_shared_bytes(data.data, data.start, data.len).map_err(convert_fst_error)),
         ReadOnlySource::Mmap(mmap_readonly) => try!(Fst::from_mmap(mmap_readonly).map_err(convert_fst_error)),
     }))
 }
@@ -141,6 +141,7 @@ mod tests {
         assert_eq!(keys.next().unwrap(), "abc".as_bytes());
         assert_eq!(keys.next().unwrap(), "abcd".as_bytes());
         assert_eq!(keys.next(), None);
+ 
     }
 
 }
diff --git a/src/directory/mod.rs b/src/directory/mod.rs
index f9bc46f7ee..d554334662 100644
--- a/src/directory/mod.rs
+++ b/src/directory/mod.rs
@@ -2,16 +2,18 @@ mod mmap_directory;
 mod ram_directory;
 mod directory;
 mod read_only_source;
+mod shared_vec_slice;
 
 use std::io::{Seek, Write};
 use std::io;
+
 use std::path::PathBuf;
+pub use self::shared_vec_slice::SharedVecSlice;
 
 pub use self::read_only_source::ReadOnlySource;
 pub use self::directory::Directory;
 pub use self::ram_directory::RAMDirectory;
 pub use self::mmap_directory::MmapDirectory;
-pub use self::ram_directory::SharedVec;
 
 
 ////////////////////////////////////////
@@ -35,6 +37,7 @@ mod tests {
 
     use super::*;
     use std::path::Path;
+    use std::io::SeekFrom;
 
     #[test]
     fn test_ram_directory() {
@@ -48,12 +51,13 @@ mod tests {
         test_directory(&mut mmap_directory);
     }
 
-    fn test_directory(directory: &mut Directory) {
+    fn test_directory_simple(directory: &mut Directory) {
         {
             let mut write_file = directory.open_write(Path::new("toto")).unwrap();
             write_file.write_all(&[4]).unwrap();
             write_file.write_all(&[3]).unwrap();
             write_file.write_all(&[7,3,5]).unwrap();
+            write_file.flush().unwrap();
         }
         let read_file = directory.open_read(Path::new("toto")).unwrap();
         let data: &[u8] = &*read_file;
@@ -65,4 +69,30 @@ mod tests {
         assert_eq!(data[4], 5);
     }
 
+
+    fn test_directory_seek(directory: &mut Directory) {
+        {
+            let mut write_file = directory.open_write(Path::new("toto_seek")).unwrap();
+            write_file.write_all(&[4]).unwrap();
+            write_file.write_all(&[3]).unwrap();
+            write_file.write_all(&[7,3,5]).unwrap();
+            write_file.seek(SeekFrom::Start(0)).unwrap();
+            write_file.write_all(&[3,1]).unwrap();
+            write_file.flush().unwrap();
+        }
+        let read_file = directory.open_read(Path::new("toto_seek")).unwrap();
+        let data: &[u8] = &*read_file;
+        assert_eq!(data.len(), 5);
+        assert_eq!(data[0], 3);
+        assert_eq!(data[1], 1);
+        assert_eq!(data[2], 7);
+        assert_eq!(data[3], 3);
+        assert_eq!(data[4], 5);
+    }
+
+    fn test_directory(directory: &mut Directory) {
+        test_directory_simple(directory);
+        test_directory_seek(directory);
+    }
+
 }
diff --git a/src/directory/ram_directory.rs b/src/directory/ram_directory.rs
index 6a3f00f40f..faae203dfd 100644
--- a/src/directory/ram_directory.rs
+++ b/src/directory/ram_directory.rs
@@ -8,40 +8,81 @@ use std::path::{Path, PathBuf};
 use directory::OpenError;
 use directory::WritePtr;
 use std::result;
+use super::SharedVecSlice;
 use Result;
 
-#[derive(Clone)]
-pub struct SharedVec(Arc<RwLock<Cursor<Vec<u8>>>>);
-
 
-pub struct RAMDirectory {
-    fs: HashMap<PathBuf, SharedVec>,
+struct VecWriter {
+    path: PathBuf,
+    shared_directory: InnerDirectory,
+    data: Cursor<Vec<u8>>,
+    is_flushed: bool,
 }
 
-impl SharedVec {
-    pub fn new() -> SharedVec {
-        SharedVec(Arc::new( RwLock::new(Cursor::new(Vec::new())) ))
+
+impl Drop for VecWriter {
+    fn drop(&mut self) {
+        if !self.is_flushed  {
+            panic!("You forgot to flush {:?} before its writter got Drop. Do not rely on drop.", self.path)
+        }
     }
-    
-    pub fn copy_vec(&self,) -> Vec<u8> {
-        self.0.read().unwrap().clone().into_inner()
+}
+
+impl Seek for VecWriter {
+    fn seek(&mut self, pos: SeekFrom) -> io::Result<u64> {
+        self.data.seek(pos)
     }
 }
 
-impl Write for SharedVec {
+impl Write for VecWriter {
     fn write(&mut self, buf: &[u8]) -> io::Result<usize> {
-        try!(self.0.write().unwrap().write(buf));
+        self.is_flushed = false;
+        try!(self.data.write(buf));
         Ok(buf.len())
     }
+
     fn flush(&mut self) -> io::Result<()> {
-        Ok(())
+        self.is_flushed = true;
+        self.shared_directory.write(self.path.clone(), self.data.get_ref())
     }
 }
 
-impl Seek for SharedVec {
-    fn seek(&mut self, pos: SeekFrom) -> io::Result<u64> {
-        self.0.write().unwrap().seek(pos)
+#[derive(Clone)]
+struct InnerDirectory(Arc<RwLock<HashMap<PathBuf, Arc<Vec<u8>>>>>);
+
+impl InnerDirectory {
+
+    fn new() -> InnerDirectory {
+        InnerDirectory(Arc::new(RwLock::new(HashMap::new())))
+    }
+
+    fn write(&self, path: PathBuf, data: &Vec<u8>) -> io::Result<()> {
+        let mut map = try!(
+            self.0
+                .write()
+                .map_err(|_| io::Error::new(io::ErrorKind::Other, format!("Failed to lock the directory, when trying to write {:?}", path)))
+        );
+        map.insert(path, Arc::new(data.clone()));
+        Ok(())
     }
+
+    fn open_read(&self, path: &Path) -> result::Result<ReadOnlySource, OpenError> { 
+        self.0
+            .read()
+            .map_err(|_| {
+                    let io_err = io::Error::new(io::ErrorKind::Other, format!("Failed to read lock for the directory, when trying to read {:?}", path));
+                    OpenError::IOError(io_err)
+            })
+            .and_then(|readable_map| {
+                readable_map
+                .get(path)
+                .ok_or_else(|| OpenError::FileDoesNotExist(PathBuf::from(path)))
+                .map(|data| {
+                    ReadOnlySource::Anonymous(SharedVecSlice::new(data.clone()))
+                })
+            })
+    }
+
 }
 
 impl fmt::Debug for RAMDirectory {
@@ -53,33 +94,37 @@ impl fmt::Debug for RAMDirectory {
 impl RAMDirectory {
     pub fn create() -> RAMDirectory {
         RAMDirectory {
-            fs: HashMap::new()
+            fs: InnerDirectory::new()
         }
     }
 }
 
+
+pub struct RAMDirectory {
+    fs: InnerDirectory,
+}
+
 impl Directory for RAMDirectory {
     fn open_read(&self, path: &Path) -> result::Result<ReadOnlySource, OpenError> {
-        match self.fs.get(path) {
-            Some(ref data) => {
-                let data_copy = data.copy_vec();
-                Ok(ReadOnlySource::Anonymous(data_copy))
-            },
-            None => {
-                Err(OpenError::FileDoesNotExist(PathBuf::from(path)))
-            }
-        }
+        self.fs.open_read(path)
     }
+    
     fn open_write(&mut self, path: &Path) -> Result<WritePtr> {
-        let full_path = PathBuf::from(&path);
-        let data = SharedVec::new();
-        self.fs.insert(full_path, data.clone());
-        Ok(Box::new(data))
+        let mut vec_writer = VecWriter {
+            path: PathBuf::from(path),
+            data: Cursor::new(Vec::new()),
+            shared_directory: self.fs.clone(),
+            is_flushed: false,
+        };
+        // force the creation of the file to mimick the MMap directory.
+        try!(vec_writer.flush());
+        Ok(Box::new(vec_writer))
     }
 
     fn atomic_write(&mut self, path: &Path, data: &[u8]) -> Result<()> {
         let mut write = try!(self.open_write(path));
         try!(write.write_all(data));
+        try!(write.flush());
         Ok(())
     }
 
diff --git a/src/directory/read_only_source.rs b/src/directory/read_only_source.rs
index 9af50c5a03..0733a35130 100644
--- a/src/directory/read_only_source.rs
+++ b/src/directory/read_only_source.rs
@@ -1,6 +1,7 @@
 use fst::raw::MmapReadOnly;
 use std::ops::Deref;
 use std::io::Cursor;
+use super::SharedVecSlice;
 
 ////////////////////////////////////////
 // Read only source.
@@ -8,7 +9,7 @@ use std::io::Cursor;
 
 pub enum ReadOnlySource {
     Mmap(MmapReadOnly),
-    Anonymous(Vec<u8>),
+    Anonymous(SharedVecSlice),
 }
 
 impl Deref for ReadOnlySource {
@@ -25,12 +26,18 @@ impl ReadOnlySource {
         self.as_slice().len()
     }
 
+    pub fn empty() -> ReadOnlySource {
+        ReadOnlySource::Anonymous(SharedVecSlice::empty())
+    }
+
     pub fn as_slice(&self,) -> &[u8] {
         match *self {
             ReadOnlySource::Mmap(ref mmap_read_only) => unsafe { 
                 mmap_read_only.as_slice()
             },
-            ReadOnlySource::Anonymous(ref shared_vec) => shared_vec.as_slice(),
+            ReadOnlySource::Anonymous(ref shared_vec) => {
+                shared_vec.as_slice()
+            },
         }
     }
 
@@ -45,8 +52,7 @@ impl ReadOnlySource {
                 ReadOnlySource::Mmap(sliced_mmap)
             }
             ReadOnlySource::Anonymous(ref shared_vec) => {
-                let sliced_data: Vec<u8> = Vec::from(&shared_vec[from_offset..to_offset]);
-                ReadOnlySource::Anonymous(sliced_data)
+                ReadOnlySource::Anonymous(shared_vec.slice(from_offset, to_offset))
             },
         }
     }
diff --git a/src/directory/shared_vec_slice.rs b/src/directory/shared_vec_slice.rs
new file mode 100644
index 0000000000..e51c993486
--- /dev/null
+++ b/src/directory/shared_vec_slice.rs
@@ -0,0 +1,36 @@
+use std::sync::Arc;
+
+#[derive(Clone)]
+pub struct SharedVecSlice {
+    pub data: Arc<Vec<u8>>,
+    pub start: usize, 
+    pub len: usize
+}
+
+impl SharedVecSlice {
+
+    pub fn empty() -> SharedVecSlice {
+        SharedVecSlice::new(Arc::new(Vec::new()))
+    }
+
+    pub fn new(data: Arc<Vec<u8>>) -> SharedVecSlice {
+        let data_len = data.len();
+        SharedVecSlice {
+            data: data,
+            start: 0,
+            len: data_len,
+        }
+    }
+
+    pub fn as_slice(&self,) -> &[u8] {
+        &self.data[self.start..self.start + self.len]
+    }
+
+    pub fn slice(&self, from_offset: usize, to_offset:usize) -> SharedVecSlice {
+        SharedVecSlice {
+            data: self.data.clone(),
+            start: self.start + from_offset,
+            len: to_offset - from_offset,
+        }
+    }
+}
diff --git a/src/fastfield/serializer.rs b/src/fastfield/serializer.rs
index ca46a84d50..e23614af5a 100644
--- a/src/fastfield/serializer.rs
+++ b/src/fastfield/serializer.rs
@@ -88,6 +88,7 @@ impl FastFieldSerializer {
         self.written_size += try!(self.fields.serialize(&mut self.write));
         try!(self.write.seek(SeekFrom::Start(0)));
         try!((header_offset as u32).serialize(&mut self.write));
+        try!(self.write.flush());
         Ok(self.written_size)
     }
 }
diff --git a/src/postings/mod.rs b/src/postings/mod.rs
index c4d55bd0fd..4272c363aa 100644
--- a/src/postings/mod.rs
+++ b/src/postings/mod.rs
@@ -52,11 +52,12 @@ mod tests {
         let mut posting_serializer = PostingsSerializer::open(&segment).unwrap();
         let term = Term::from_field_text(text_field, "abc");
         posting_serializer.new_term(&term, 3).unwrap();
-        for _ in 0..3 {
-            let a = vec!(1,2,3,2);
-            posting_serializer.write_doc(0, 2, &a).unwrap();
+        for doc_id in 0u32..3u32 {
+            let positions = vec!(1,2,3,2);
+            posting_serializer.write_doc(doc_id, 2, &positions).unwrap();
         }
         posting_serializer.close_term().unwrap();
+        posting_serializer.close().unwrap();
         let read = segment.open_read(SegmentComponent::POSITIONS).unwrap();
         assert_eq!(read.len(), 13);
     }
diff --git a/src/postings/serializer.rs b/src/postings/serializer.rs
index 3da98e4d6f..83cf8ac8b2 100644
--- a/src/postings/serializer.rs
+++ b/src/postings/serializer.rs
@@ -31,6 +31,7 @@ pub struct PostingsSerializer {
     position_deltas: Vec<u32>,
     schema: Schema,
     text_indexing_options: TextIndexingOptions,
+    term_open: bool,
 }
 
 impl PostingsSerializer {
@@ -55,6 +56,7 @@ impl PostingsSerializer {
             position_deltas: Vec::new(),
             schema: schema,
             text_indexing_options: TextIndexingOptions::Unindexed,
+            term_open: false,
         })
     }
 
@@ -76,7 +78,10 @@ impl PostingsSerializer {
     }
 
     pub fn new_term(&mut self, term: &Term, doc_freq: DocId) -> io::Result<()> {
-        try!(self.close_term());
+        if self.term_open {
+            panic!("Called new_term, while the previous term was not closed.");
+        }
+        self.term_open = true;
         self.load_indexing_options(term.get_field());
         self.doc_ids.clear();
         self.last_doc_id_encoded = 0;
@@ -92,14 +97,22 @@ impl PostingsSerializer {
     }
 
     pub fn close_term(&mut self,) -> io::Result<()> {
-        if !self.doc_ids.is_empty() {
-            {
-                let block_encoded = self.block_encoder.compress_vint_sorted(&self.doc_ids, self.last_doc_id_encoded);
-                self.written_bytes_postings += block_encoded.len();
-                try!(self.postings_write.write_all(block_encoded));
-            }
-            if self.text_indexing_options.is_termfreq_enabled() {
+        if self.term_open {
+            if !self.doc_ids.is_empty() {
+                // we have doc ids waiting to be written
+                // this happens when the number of doc ids is 
+                // not a perfect multiple of our block size.
+                //
+                // In that case, the remaining part is encoded
+                // using variable int encoding.
                 {
+                    let block_encoded = self.block_encoder.compress_vint_sorted(&self.doc_ids, self.last_doc_id_encoded);
+                    self.written_bytes_postings += block_encoded.len();
+                    try!(self.postings_write.write_all(block_encoded));
+                    self.doc_ids.clear();
+                }
+                // ... Idem for term frequencies 
+                if self.text_indexing_options.is_termfreq_enabled() {
                     let block_encoded = self.block_encoder.compress_vint_unsorted(&self.term_freqs[..]);
                     for num in block_encoded {
                         self.written_bytes_postings += try!(num.serialize(&mut self.postings_write));
@@ -107,14 +120,16 @@ impl PostingsSerializer {
                     self.term_freqs.clear();
                 }
             }
-            self.doc_ids.clear();
-        }
-        if self.text_indexing_options.is_position_enabled() {
-            self.written_bytes_positions += try!(VInt(self.position_deltas.len() as u64).serialize(&mut self.positions_write));
-            let positions_encoded: &[u8] = self.positions_encoder.compress_unsorted(&self.position_deltas[..]);
-            try!(self.positions_write.write_all(positions_encoded));
-            self.written_bytes_positions += positions_encoded.len();
-            self.position_deltas.clear();
+            // On the other hand, positions are entirely buffered until the
+            // end of the term, at which point they are compressed and written.
+            if self.text_indexing_options.is_position_enabled() {
+                self.written_bytes_positions += try!(VInt(self.position_deltas.len() as u64).serialize(&mut self.positions_write));
+                let positions_encoded: &[u8] = self.positions_encoder.compress_unsorted(&self.position_deltas[..]);
+                try!(self.positions_write.write_all(positions_encoded));
+                self.written_bytes_positions += positions_encoded.len();
+                self.position_deltas.clear();
+            }
+            self.term_open = false;
         }
         Ok(())
     }
diff --git a/src/postings/writer.rs b/src/postings/writer.rs
index 3293009e4b..1726b9c10e 100644
--- a/src/postings/writer.rs
+++ b/src/postings/writer.rs
@@ -119,6 +119,7 @@ impl<Rec: Recorder + 'static> PostingsWriter for SpecializedPostingsWriter<Rec>
             let term_docfreq = term_postings_writer.doc_freq();
             try!(serializer.new_term(term, term_docfreq));
             try!(term_postings_writer.serialize(serializer));
+            try!(serializer.close_term());
         }
         Ok(())
     }
diff --git a/src/query/daat_multiterm_scorer.rs b/src/query/daat_multiterm_scorer.rs
index 65a6ba78aa..798942e4a3 100644
--- a/src/query/daat_multiterm_scorer.rs
+++ b/src/query/daat_multiterm_scorer.rs
@@ -204,9 +204,10 @@ mod tests {
     use postings::{DocSet, VecPostings};
     use query::TfIdf;
     use query::Scorer;
-    use directory::ReadOnlySource;
-    use directory::SharedVec;
+    use directory::Directory;
+    use directory::RAMDirectory;
     use schema::Field;
+    use std::path::Path;
     use query::Occur;
     use fastfield::{U32FastFieldReader, U32FastFieldWriter, FastFieldSerializer};
 
@@ -216,12 +217,16 @@ mod tests {
         for val in vals {
             u32_field_writer.add_val(val);
         }
-        let data = SharedVec::new();
-        let write: Box<SharedVec> = Box::new(data.clone());
-        let mut serializer = FastFieldSerializer::new(write).unwrap();
-        u32_field_writer.serialize(&mut serializer).unwrap();
-        serializer.close().unwrap();
-        U32FastFieldReader::open(ReadOnlySource::Anonymous(data.copy_vec())).unwrap()
+        let path = Path::new("some_path");
+        let mut directory = RAMDirectory::create();
+        {
+            let write = directory.open_write(&path).unwrap();
+            let mut serializer = FastFieldSerializer::new(write).unwrap();
+            u32_field_writer.serialize(&mut serializer).unwrap();
+            serializer.close().unwrap();
+        }
+        let read = directory.open_read(&path).unwrap();
+        U32FastFieldReader::open(read).unwrap()
     }
     
     fn abs_diff(left: f32, right: f32) -> f32 {