diff --git a/src/main.rs b/src/main.rs
index aca3fa2..adeea78 100644
--- a/src/main.rs
+++ b/src/main.rs
@@ -1,3 +1,5 @@
+#![allow(unused)]
+
 #[derive(PartialEq, Debug, Clone, Copy)]
 struct Range {
     // todo: consider just using a map instead
@@ -15,11 +17,11 @@ enum RangeLocation {
 }
 
 impl Range {
-    pub fn new(start: usize, end: usize) -> Self {
+    fn new(start: usize, end: usize) -> Self {
         Self { start, end }
     }
 
-    pub fn where_in_range(&self, index: usize) -> RangeLocation {
+    fn where_in_range(&self, index: usize) -> RangeLocation {
         if index == self.start {
             return RangeLocation::LeftEnd;
         }
@@ -32,25 +34,30 @@ impl Range {
         RangeLocation::Middle
     }
 
-    pub fn is_empty(&self) -> bool {
+    fn contains(&self, range: &Range) -> bool {
+        range.start >= self.start && range.end <= self.end
+    }
+
+    fn is_empty(&self) -> bool {
         self.start >= self.end
     }
 }
 
 pub struct Blob {
+    // todo: consider fallible_vec to detect OOM on append
     bytes: Vec<u8>,
     valid_ranges: Vec<Range>,
 }
 
 impl Blob {
-    pub fn new() -> Self {
+    fn new() -> Self {
         Self {
             bytes: Vec::new(),
             valid_ranges: Vec::new(),
         }
     }
 
-    pub fn append(&mut self, input: &[u8]) {
+    fn append(&mut self, input: &[u8]) {
         let new_range = Range {
             start: self.bytes.len(),
             end: self.bytes.len() + input.len(),
@@ -67,8 +74,33 @@ impl Blob {
     }
 
     // todo: better way to do the return type?
-    pub fn read(&mut self, start: usize, len: usize, output: &mut [u8]) {
+    fn read(&mut self, start: usize, len: usize, output: &mut [u8]) -> Result<(), BlobError> {
         assert!(len == output.len()); // todo: add error handling
+
+        let read_range = Range::new(start, start + len);
+
+        // check read within bytes len
+        match self.valid_ranges.last() {
+            Some(last) if read_range.end <= last.end => {}
+            _ => return Err(BlobError::InvalidRange),
+        }
+
+        // check bytes are not already read
+        let mut contains = false;
+        for valid_range in self.valid_ranges.iter() {
+            if valid_range.contains(&read_range) {
+                contains = true;
+                break;
+            }
+            if read_range.start >= valid_range.end {
+                return Err(BlobError::BytesAlreadyRead);
+            }
+        }
+        if !contains {
+            // Invalid range already caught above
+            return Err(BlobError::BytesAlreadyRead);
+        }
+
         let slice = &mut self.bytes[start..start + len];
         slice.swap_with_slice(output);
 
@@ -106,11 +138,65 @@ impl Blob {
             i += 1;
         }
         self.valid_ranges = new_ranges;
+        Ok(())
+    }
+}
+
+// combine all errors into one, since there aren't that many
+#[derive(Debug, PartialEq)]
+enum BlobError {
+    BlobDNE,
+    BlobExists,
+    InvalidRange,
+    BytesAlreadyRead,
+}
+
+struct BlobManager {
+    // fine if we assume:
+    // * IDs are sequentially generated
+    // * and IDs can be re-used
+    blobs: Vec<Option<Blob>>,
+}
+
+impl BlobManager {
+    pub fn new() -> Self {
+        Self { blobs: Vec::new() }
+    }
+
+    // todo: is usize appropriate datatype for ID?
+    pub fn create_blob(&mut self, id: usize) -> Result<(), BlobError> {
+        if let Some(Some(_)) = self.blobs.get_mut(id) {
+            return Err(BlobError::BlobExists);
+        } else if id >= self.blobs.len() {
+            // todo: better strategy possible?
+            self.blobs.resize_with(id + 1, || None);
+        }
+        self.blobs[id] = Some(Blob::new());
+        Ok(())
+    }
+
+    pub fn append(&mut self, id: usize, input: &[u8]) -> Result<(), BlobError> {
+        if let Some(Some(b)) = self.blobs.get_mut(id) {
+            b.append(input);
+            return Ok(());
+        }
+        return Err(BlobError::BlobDNE);
+    }
+
+    pub fn read(&mut self, id: usize, start: usize, len: usize) -> Result<Vec<u8>, BlobError> {
+        if let Some(Some(b)) = self.blobs.get_mut(id) {
+            let mut read = vec![0; len];
+            b.read(start, len, &mut read)?; // propagate error to caller
+            // todo: if read clears a blob entirely should the blob be removed?
+            Ok(read)
+        } else {
+            Err(BlobError::BlobDNE)
+        }
     }
 }
 
 fn main() {
-    println!("run cargo test instead")
+    println!("run cargo test instead");
 }
 
 #[cfg(test)]
@@ -129,6 +215,21 @@ mod tests {
         assert_eq!(range.where_in_range(1), RangeLocation::Middle);
         assert_eq!(range.where_in_range(2), RangeLocation::RightEnd);
         assert_eq!(range.where_in_range(3), RangeLocation::No);
+
+        assert!(range.contains(&range2));
+        assert!(range2.contains(&range));
+
+        let range3 = Range::new(0, 1);
+        assert!(range.contains(&range3));
+        assert!(!range3.contains(&range));
+
+        let range4 = Range::new(2, 3);
+        assert!(range.contains(&range4));
+        assert!(!range4.contains(&range));
+
+        let range5 = Range::new(2, 4);
+        assert!(!range.contains(&range5));
+        assert!(!range5.contains(&range));
     }
 
     #[test]
@@ -154,6 +255,26 @@ mod tests {
         let mut read = vec![0; 1];
         b.read(4, 1, &mut read); // middle read
         assert_eq!(b.valid_ranges, &[Range::new(3, 4), Range::new(5, 6)]);
-        println!("blob: {:?}", b.bytes);
+    }
+
+    #[test]
+    fn test_blob_manager() {
+        let mut bm = BlobManager::new();
+        bm.create_blob(0);
+        bm.append(0, &[b'a', b'b', b'c']);
+        let v = Vec::from([b'a', b'b']);
+        assert_eq!(bm.read(0, 0, 2).unwrap(), v);
+
+        bm.create_blob(1);
+        bm.append(1, &[b'a', b'b', b'c']);
+        assert_eq!(bm.read(1, 0, 4).unwrap_err(), BlobError::InvalidRange);
+
+        let read = bm.read(1, 0, 2).unwrap();
+        assert_eq!(read, v);
+        assert_eq!(bm.read(1, 0, 2).unwrap_err(), BlobError::BytesAlreadyRead);
+        assert_eq!(bm.read(1, 1, 2).unwrap_err(), BlobError::BytesAlreadyRead);
+
+        assert_eq!(bm.read(2, 0, 2).unwrap_err(), BlobError::BlobDNE);
+        assert_eq!(bm.create_blob(1).unwrap_err(), BlobError::BlobExists);
     }
 }