The VarInt Specification

To correctly parse a Bitcoin transaction, you must follow the strict thresholds defined in the VarInt (CompactSize) specification. The logic is simple but requires careful byte-reading.

1. The Threshold Rules

When a parser encounters a VarInt field, it reads the first byte and follows this logic:

First Byte	Meaning	Data to Read Next
< 0xFD	The value is the byte itself.	0 Bytes
== 0xFD	The value is a 16-bit integer.	2 Bytes (Little Endian)
== 0xFE	The value is a 32-bit integer.	4 Bytes (Little Endian)
== 0xFF	The value is a 64-bit integer.	8 Bytes (Little Endian)

2. Little Endian Requirement

Except for the 1-byte case, all subsequent bytes must be read in Little Endian order.

• Example: A script length of 512 bytes.

• 512 in hex is 0x0200.

• Serialized: FD 00 02 (Marker FD + LE bytes 00 02).

3. Handling Zero

The number zero is a valid VarInt. It is represented as a single byte: 0x00. You will see this in transactions with zero outputs (rare) or empty scripts.

4. Avoiding Redundancy

A value like 10 could technically be represented as FD 0A 00 (the 3-byte version). However, the Bitcoin consensus rules require Canonical Encoding.

• If a number can fit in 1 byte, it must be 1 byte.

• If you use a 3-byte VarInt to store a 1-byte value, your transaction is "Non-Standard" and will be rejected by most nodes.

5. Security: The "OOM" Attack

Parsers must be careful when reading the output count VarInt.

• If an attacker sends a transaction claiming to have 4 billion outputs (FE FF FF FF FF), a poorly written node might try to allocate gigabytes of RAM to store those outputs before realizing they don't actually exist in the packet.

• Modern nodes verify the packet size matches the claimed VarInt count before allocating memory.

Decimal Value	VarInt Hex	Total Size
252	FC	1 Byte
253	FD FD 00	3 Bytes
65,535	FD FF FF	3 Bytes
65,536	FE 00 00 01 00	5 Bytes

In the next section, we will compare VarInt vs. Fixed-Width Integers.