Routers ​

The routers are the most important features for Hono.

Hono has five routers.

RegExpRouter ​

RegExpRouter is the fastest router in the JavaScript world.

Although this is called "RegExp" it is not an Express-like implementation using path-to-regexp. They are using linear loops. Therefore, regular expression matching will be performed for all routes and the performance will be degraded as you have more routes.

Hono's RegExpRouter turns the route pattern into "one large regular expression". Then it can get the result with one-time matching.

This works faster than methods that use tree-based algorithms such as radix-tree in most cases.

TrieRouter ​

TrieRouter is the router using the Trie-tree algorithm. It does not use liner loops as same as RegExpRouter.

This router is not as fast as the RegExpRouter, but it is much faster than the Express router. TrieRouter supports all patterns though RegExpRouter does not.

SmartRouter ​

RegExpRouter doesn't support all routing patterns. Therefore, it's usually used in combination with another router that does support all patterns.

SmartRouter will select the best router by inferring from the registered routers. Hono uses SmartRouter and the two routers by default:

// Inside the core of Hono.
readonly defaultRouter: Router = new SmartRouter({
  routers: [new RegExpRouter(), new TrieRouter()],
})

// Inside the core of Hono.
readonly defaultRouter: Router = new SmartRouter({
  routers: [new RegExpRouter(), new TrieRouter()],
})

When the application starts, SmartRouter detects the fastest router based on routing and continues to use it.

LinearRouter ​

RegExpRouter is fast, but the route registration phase can be slightly slow. So, it's not suitable for an environment that initializes with every request.

LinearRouter is optimized for "one shot" situations. Route registration is significantly faster than with RegExpRouter because it adds the route without compiling strings, using a linear approach.

The following is one of the benchmark results, which includes the route registration phase.

• GET /user/lookup/username/hey
----------------------------------------------------- -----------------------------
LinearRouter     1.82 µs/iter      (1.7 µs … 2.04 µs)   1.84 µs   2.04 µs   2.04 µs
MedleyRouter     4.44 µs/iter     (4.34 µs … 4.54 µs)   4.48 µs   4.54 µs   4.54 µs
FindMyWay       60.36 µs/iter      (45.5 µs … 1.9 ms)  59.88 µs  78.13 µs  82.92 µs
KoaTreeRouter    3.81 µs/iter     (3.73 µs … 3.87 µs)   3.84 µs   3.87 µs   3.87 µs
TrekRouter       5.84 µs/iter     (5.75 µs … 6.04 µs)   5.86 µs   6.04 µs   6.04 µs

summary for GET /user/lookup/username/hey
  LinearRouter
   2.1x faster than KoaTreeRouter
   2.45x faster than MedleyRouter
   3.21x faster than TrekRouter
   33.24x faster than FindMyWay

• GET /user/lookup/username/hey
----------------------------------------------------- -----------------------------
LinearRouter     1.82 µs/iter      (1.7 µs … 2.04 µs)   1.84 µs   2.04 µs   2.04 µs
MedleyRouter     4.44 µs/iter     (4.34 µs … 4.54 µs)   4.48 µs   4.54 µs   4.54 µs
FindMyWay       60.36 µs/iter      (45.5 µs … 1.9 ms)  59.88 µs  78.13 µs  82.92 µs
KoaTreeRouter    3.81 µs/iter     (3.73 µs … 3.87 µs)   3.84 µs   3.87 µs   3.87 µs
TrekRouter       5.84 µs/iter     (5.75 µs … 6.04 µs)   5.86 µs   6.04 µs   6.04 µs

summary for GET /user/lookup/username/hey
  LinearRouter
   2.1x faster than KoaTreeRouter
   2.45x faster than MedleyRouter
   3.21x faster than TrekRouter
   33.24x faster than FindMyWay

For situations like Fastly Compute@Edge, it's better to use LinearRouter with the hono/quick preset.

PatternRouter ​

PatternRouter is the smallest router among Hono's routers.

While Hono is already compact, if you need to make it even smaller for an environment with limited resources, you can use PatternRouter.

An application using only PatternRouter is under 12KB in size.

$ npx wrangler dev --minify ./src/index.ts
 ⛅️ wrangler 2.20.0
--------------------
⬣ Listening at http://0.0.0.0:8787
- http://127.0.0.1:8787
- http://192.168.128.165:8787
Total Upload: 11.47 KiB / gzip: 4.34 KiB

$ npx wrangler dev --minify ./src/index.ts
 ⛅️ wrangler 2.20.0
--------------------
⬣ Listening at http://0.0.0.0:8787
- http://127.0.0.1:8787
- http://192.168.128.165:8787
Total Upload: 11.47 KiB / gzip: 4.34 KiB