m-chrzan.xyz
aboutsummaryrefslogtreecommitdiff
diff options
context:
space:
mode:
authorMarcin Chrzanowski <marcin.j.chrzanowski@gmail.com>2019-07-31 22:36:24 -0700
committerMarcin Chrzanowski <marcin.j.chrzanowski@gmail.com>2019-07-31 22:36:24 -0700
commita8bd31af4d8531f4bd0d7d7b088305dc7cab70f8 (patch)
tree010d35a45a038ab7e5e4236179c94e5ec7daff84
parent77084cd3f4567f923ed236ddb273613e8e5c3397 (diff)
Publish Polymorphism in Solidity post
-rw-r--r--src/blog/polymorphism-in-solidity.html248
1 files changed, 248 insertions, 0 deletions
diff --git a/src/blog/polymorphism-in-solidity.html b/src/blog/polymorphism-in-solidity.html
new file mode 100644
index 0000000..a6a6aca
--- /dev/null
+++ b/src/blog/polymorphism-in-solidity.html
@@ -0,0 +1,248 @@
+title: Polymorphism in Solidity
+date: July 31, 16:44
+---
+<p>
+Solidity is in many ways similar to C. It's a low-level language, sitting just a
+thin layer of abstraction above its underlying bytecode. Just like C, it lacks
+some convenient mechanisms from higher level languages.
+</p>
+
+<p>
+There's a cool method for implementing simple object-orientation (complete with
+polymorphism) in C that can also be applied in Solidity to solve similar
+problems.
+</p>
+
+<p>
+Let's look at how Linux kernel programmers deal with filesystems.
+</p>
+
+<p>
+Each filesystem needs its own low-level implementation. At the same time, it
+would be nice to have the abstract concept of a <em>file</em>, and be able to
+write generic code that can interact with <em>files</em> living on any sort of
+filesystem. Sounds like polymorphism.
+</p>
+
+<p>
+Here's the first few lines of the definition of
+<a href='https://elixir.bootlin.com/linux/v5.2.4/source/include/linux/fs.h#L922'>
+ <code>struct file</code>
+</a>, used
+to keep track of a file in the Linux kernel.
+
+<pre>
+struct file {
+ struct path f_path;
+ struct inode *f_inode;
+ <b>const struct file_operations *f_op</b>;
+ // ...
+};
+</pre>
+</a>
+
+<p>
+The important bit is the <code>f_op</code> field, of type <code>struct
+ file_operations</code>. Let's
+look at (an abridged version of)
+<a href='https://elixir.bootlin.com/linux/v5.2.4/source/include/linux/fs.h#L1791'>
+ <em>its</em> definition
+</a>.
+
+<pre>
+struct file_operations {
+ struct module *owner;
+ loff_t (*<b>llseek</b>) (struct file *, loff_t, int);
+ ssize_t (*<b>read</b>) (struct file *, char __user *, size_t, loff_t *);
+ ssize_t (*<b>write</b>) (struct file *, const char __user *, size_t, loff_t *);
+ ssize_t (*<b>read_iter</b>) (struct kiocb *, struct iov_iter *);
+ ssize_t (*<b>write_iter</b>) (struct kiocb *, struct iov_iter *);
+ int (*<b>mmap</b>) (struct file *, struct vm_area_struct *);
+ int (*<b>open</b>) (struct inode *, struct file *);
+ // ...
+};
+</pre>
+</a>
+
+It's mostly a bunch of function pointers that appear to be... operations one
+might want to perform on a file.
+</p>
+
+<p>
+To emulate OO, inside our "object" (<code>struct file</code>) we manually store
+a container for its "methods" (<code>struct file_operations</code>). Each of
+these, as its first argument, takes a pointer to a <code>struct file</code> that
+it's going to operate on.
+</p>
+
+<p>
+With this in place, we can now define a generic
+<a href='https://elixir.bootlin.com/linux/v5.2.4/source/fs/read_write.c#L421'>
+ <code>read</code>
+</a> system call:
+
+<pre>
+ssize_t __vfs_read(struct file *file, char __user *buf, size_t count,
+ loff_t *pos)
+{
+ if (<b>file-&gt;f_op-&gt;read</b>)
+ return <b>file-&gt;f_op-&gt;read(</b>file, buf, count, pos<b>)</b>;
+ else if (<b>file-&gt;f_op-&gt;read_iter</b>)
+ return new_sync_read(file, buf, count, pos);
+ else
+ return -EINVAL;
+}
+</pre>
+
+On the other hand, the file
+<a href='https://elixir.bootlin.com/linux/v5.2.4/source/fs/ext4/file.c#L508'>
+ <code>fs/ext4/file.c</code>
+</a> defines operations specific to the ext4 file system and a
+<code>file_operations</code> struct:
+
+<pre>
+const struct file_operations ext4_file_operations = {
+ .llseek = ext4_llseek,
+ .read_iter = ext4_file_read_iter,
+ .write_iter = ext4_file_write_iter,
+ // ...
+};
+</pre>
+</p>
+
+<p>
+We can do the same thing in Solidity!
+</p>
+
+<p>
+The example we'll work with is a decentralized exchange. Users can call a
+<code>trade</code> function with the following signature:
+
+<pre>
+function trade(address sellCurrency, address buyCurrency, uint256 sellAmount);
+</pre>
+
+They specify a currency pair (<code>sellCurrency</code> and <code>buyCurrency</code> -
+the currency they're selling to and buying from the exchange) and the amount of
+<code>sellCurrency</code> they're giving to the exchange. The smart contract then
+calculates the amount of <code>buyCurrency</code> the user should receive and
+transfers that to them.
+</p>
+
+<p>
+To compilcate things a little, let's say that the exchange deals with more than
+just ERC20 tokens. Let's allow for ERC20 - ERC20, ERC20 - Ether, and Ether -
+ERC20 trades.
+</p>
+
+<p>
+Here's what a first attempt at implementing this might look like:
+
+<pre>
+// Let address(0) denote Ether
+function trade(address sellCurrency, address buyCurrency, uint256 sellAmount) {
+ uint256 buyAmount = calculateBuyAmount(sellCurrency, buyCurrency, sellAmount);
+
+ // <b>take the user's sellCurrency</b>
+ if (sellCurrency == address(0)) {
+ require(msg.value == sellAmount);
+ } else {
+ ERC20(sellCurrency).transferFrom(msg.sender, address(this), sellAmount);
+ }
+
+ // <b>give the user their new buyCurrency</b>
+ if (buyCurrency == address(0)) {
+ msg.sender.transfer(buyAmount);
+ } else {
+ ERC20(buyCurrency).transfer(msg.sender, buyAmount);
+ }
+}
+</pre>
+</p>
+
+<p>
+This doesn't look terrible yet.
+</p>
+
+<p>
+Now imagine that you wanted to handle even more asset classes.
+
+<p>
+What if there was a token <code>YourToken</code> that had <code>mint</code> and
+<code>burn</code> functions callable by the exchange contract? Instead of
+holding a balance of <code>YourToken</code> you just want to either take tokens
+out of ciruclation when they're sold, or mint new ones into existence when
+they're bought.
+</p>
+
+<p>
+Or you want to support <code>MyToken</code> which I annoyingly implemented
+without following the ERC20 standard and function names differ from other tokens.
+</p>
+
+<p>
+With more and more asset classes, the complexity of the code above would
+increase.
+</p>
+
+<p>
+Now let's try to implement the same logic but taking inspiration from the Linux
+kernel's generic handling of files.
+</p>
+
+<p>
+First, let's declare the struct that will hold a currency's information and methods
+for interacting with it. This corresponds to <code>struct file</code>:
+
+<pre>
+struct Currency {
+ function (Currency, uint256) take;
+ function (Currency, uint256) give;
+ address currencyAddress;
+}
+</pre>
+</p>
+
+<p>
+Now let's implement taking and giving tokens for two different asset classes.
+
+<pre>
+function <b>ethTake</b>(Currency currencyS, uint256 amount) {
+ require(msg.value == sellAmount);
+}
+
+function <b>ethGive</b>(Currency currencyS, uint256 amount) {
+ msg.sender.transfer(buyAmount);
+}
+
+function <b>erc20Take</b>(Currency currencyS, uint256 amount) {
+ ERC20 token = ERC20(currencyS.currencyAddress);
+ token.transferFrom(msg.sender, address(this), amount);
+}
+
+function <b>erc20Give</b>(Currency currencyS, uint256 amount) {
+ ERC20 token = ERC20(currencyS.currencyAddress);
+ token.transfer(msg.sender, amount);
+}
+</pre>
+</p>
+
+<p>
+Finally, we can perform generic operations on currencies:
+
+<pre>
+function trade(Currency sellCurrency, Currency buyCurrency, uint256 sellAmount
+) {
+ uint256 buyAmount = calculateBuyAmount(sellCurrency, buyCurrency, sellAmount);
+
+ <b>sellCurrency.take(</b>sellCurrency, sellAmount<b>)</b>;
+ <b>buyCurrency.give(</b>buyCurrency, buyAmount<b>)</b>;
+}
+</pre>
+</p>
+
+<p>
+Adding support for a new asset class is now as simple as defining a pair of take/give
+functions. The code inside of <code>trade</code> need never be touched again,
+following the Open/Closed principle.
+</p>