chore: add functional options

main
sudo pacman -Syu 2022-07-12 16:55:25 +07:00
parent 20711ae4f6
commit bd0b5f9fa0
No known key found for this signature in database
GPG Key ID: D6CB5C6C567C47B0
2 changed files with 82 additions and 1 deletions

View File

@ -20,4 +20,35 @@ internal
| | | service.go
| | | repository.go
| | | models.go
</code></pre><p>All business codes are inside <code>internal</code>.<br>Each business has a different directory (<code>business_1</code>, <code>business_2</code>).<p>Inside each business, there are 2 handlers: <code>http</code>, <code>grpc</code>:<ul><li><code>http</code> is for public APIs (Android, iOS,... are clients).<li><code>grpc</code> is for internal APIs (other services are clients).</ul><p>Inside each handler, there are usually 3 layers: <code>handler</code>, <code>service</code>, <code>repository</code>:<ul><li><code>handler</code> interacts directly with gRPC or REST using specific codes (cookies,...)<li><code>service</code> is where we write business/logic codes, and only business/logic codes is written here.<li><code>repository</code> is where we write codes which interacts with database/cache like MySQL, Redis, ...</ul><p><code>handler</code> must exist inside <code>grpc</code>, <code>http</code>.<br>But <code>service</code>, <code>repository</code>, <code>models</code> can exist directly inside <code>business</code> if both <code>grpc</code>, <code>http</code> has same business/logic.<h2>Do not repeat!</h2><p>If we have too many services, some of the logic will be overlapped.<p>For example, service A and service B both need to make POST call API to service C.<br>If service A and service B both have libs to call service C to do that API, we need to move the libs to some common pkg libs.<br>So in the future, service D which needs to call C will not need to copy libs to handle service C api but only need to import from common pkg libs.<p>Another bad practice is adapter service.<br>No need to write a new service if what we need is just common pkg libs.<h2>External libs</h2><h3>Don't use cli libs (<a href=https://github.com/spf13/cobra>spf13/cobra</a>, <a href=https://github.com/urfave/cli>urfave/cli</a>) just for Go service</h3><p>What is the point to pass many params (<code>--abc</code>, <code>--xyz</code>) when what we only need is start service?<p>In my case, service starts with only config, and config should be read from file or environment like <a href=https://12factor.net/>The Twelve Factors</a> guide.<h3>Don't use <a href=https://github.com/grpc-ecosystem/grpc-gateway>grpc-ecosystem/grpc-gateway</a></h3><p>Just don't.<p>Use <a href=https://github.com/protocolbuffers/protobuf-go>protocolbuffers/protobuf-go</a>, <a href=https://github.com/grpc/grpc-go>grpc/grpc-go</a> for gRPC.<p>Write 1 for both gRPC, REST sounds good, but in the end, it is not worth it.<h3>Don't use <a href=https://github.com/uber/prototool>uber/prototool</a>, use <a href=https://github.com/bufbuild/buf>bufbuild/buf</a></h3><p>prototool is deprecated, and buf can generate, lint, format as good as prototool.<h3>Use <a href=https://github.com/gin-gonic/gin>gin-gonic/gin</a> for REST.</h3><p>Don't use <code>gin.Context</code> when pass context from handler layer to service layer, use <code>gin.Context.Request.Context()</code> instead.<h3>If you want log, just use <a href=https://github.com/uber-go/zap>uber-go/zap</a></h3><p>It is fast!<ul><li><p>Don't overuse <code>func (*Logger) With</code>. Because if log line is too long, there is a possibility that we can lost it.<li><p>Use <code>MarshalLogObject</code> when we need to hide some field of object when log (field has long or sensitive value)<li><p>Don't use <code>Panic</code>. Use <code>Fatal</code> for errors when start service to check dependencies. If you really need panic level, use <code>DPanic</code>.<li><p>Use <code>contextID</code> or <code>traceID</code> in every log lines for easily debug.</ul><h3>Don't overuse ORM libs, no need to handle another layer above SQL.</h3><p>Each ORM libs has each different syntax.<br>To learn and use those libs correctly is time consuming.<br>So just stick to plain SQL.<br>It is easier to debug when something is wrong.<p>But <code>database/sql</code> has its own limit.<br>For example, it is hard to get primary key after insert/update.<br>So may be you want to use ORM for those cases.<h3>If you want test, just use <a href=https://github.com/stretchr/testify>stretchr/testify</a>.</h3><p>It is easy to write a suite test, thanks to testify.<br>Also, for mocking, there are many options out there.<br>Pick 1 then sleep peacefully.<h3>Replace <code>go fmt</code>, <code>goimports</code> with <a href=https://github.com/mvdan/gofumpt>mvdan/gofumpt</a>.</h3><p><code>gofumpt</code> provides more rules when format Go codes.<h3>Use <a href=https://github.com/golangci/golangci-lint>golangci/golangci-lint</a>.</h3><p>No need to say more.<br>Lint or get the f out!
</code></pre><p>All business codes are inside <code>internal</code>.<br>Each business has a different directory (<code>business_1</code>, <code>business_2</code>).<p>Inside each business, there are 2 handlers: <code>http</code>, <code>grpc</code>:<ul><li><code>http</code> is for public APIs (Android, iOS,... are clients).<li><code>grpc</code> is for internal APIs (other services are clients).</ul><p>Inside each handler, there are usually 3 layers: <code>handler</code>, <code>service</code>, <code>repository</code>:<ul><li><code>handler</code> interacts directly with gRPC or REST using specific codes (cookies,...)<li><code>service</code> is where we write business/logic codes, and only business/logic codes is written here.<li><code>repository</code> is where we write codes which interacts with database/cache like MySQL, Redis, ...</ul><p><code>handler</code> must exist inside <code>grpc</code>, <code>http</code>.<br>But <code>service</code>, <code>repository</code>, <code>models</code> can exist directly inside <code>business</code> if both <code>grpc</code>, <code>http</code> has same business/logic.<h2>Do not repeat!</h2><p>If we have too many services, some of the logic will be overlapped.<p>For example, service A and service B both need to make POST call API to service C.<br>If service A and service B both have libs to call service C to do that API, we need to move the libs to some common pkg libs.<br>So in the future, service D which needs to call C will not need to copy libs to handle service C api but only need to import from common pkg libs.<p>Another bad practice is adapter service.<br>No need to write a new service if what we need is just common pkg libs.<h2>Taste on style guide</h2><h3>Use functional options, but don't overuse it!</h3><p>For simple struct with 1 or 2 fields, no need to use functional options.<p><a href=https://go.dev/play/p/0XnOLiHuoz3>Example</a>:<pre><code class=language-go>func main() {
s := NewS(WithA(1), WithB(&quot;b&quot;))
fmt.Printf(&quot;%+v\n&quot;, s)
}
type S struct {
fieldA int
fieldB string
}
type OptionS func(s *S)
func WithA(a int) OptionS {
return func(s *S) {
s.fieldA = a
}
}
func WithB(b string) OptionS {
return func(s *S) {
s.fieldB = b
}
}
func NewS(opts ...OptionS) *S {
s := &amp;S{}
for _, opt := range opts {
opt(s)
}
return s
}
</code></pre><p>In above example, I construct <code>s</code> with <code>WithA</code> and <code>WithB</code> option.<br>No need to pass direct field inside <code>s</code>.<h2>External libs</h2><h3>Don't use cli libs (<a href=https://github.com/spf13/cobra>spf13/cobra</a>, <a href=https://github.com/urfave/cli>urfave/cli</a>) just for Go service</h3><p>What is the point to pass many params (<code>--abc</code>, <code>--xyz</code>) when what we only need is start service?<p>In my case, service starts with only config, and config should be read from file or environment like <a href=https://12factor.net/>The Twelve Factors</a> guide.<h3>Don't use <a href=https://github.com/grpc-ecosystem/grpc-gateway>grpc-ecosystem/grpc-gateway</a></h3><p>Just don't.<p>Use <a href=https://github.com/protocolbuffers/protobuf-go>protocolbuffers/protobuf-go</a>, <a href=https://github.com/grpc/grpc-go>grpc/grpc-go</a> for gRPC.<p>Write 1 for both gRPC, REST sounds good, but in the end, it is not worth it.<h3>Don't use <a href=https://github.com/uber/prototool>uber/prototool</a>, use <a href=https://github.com/bufbuild/buf>bufbuild/buf</a></h3><p>prototool is deprecated, and buf can generate, lint, format as good as prototool.<h3>Use <a href=https://github.com/gin-gonic/gin>gin-gonic/gin</a> for REST.</h3><p>Don't use <code>gin.Context</code> when pass context from handler layer to service layer, use <code>gin.Context.Request.Context()</code> instead.<h3>If you want log, just use <a href=https://github.com/uber-go/zap>uber-go/zap</a></h3><p>It is fast!<ul><li><p>Don't overuse <code>func (*Logger) With</code>. Because if log line is too long, there is a possibility that we can lost it.<li><p>Use <code>MarshalLogObject</code> when we need to hide some field of object when log (field has long or sensitive value)<li><p>Don't use <code>Panic</code>. Use <code>Fatal</code> for errors when start service to check dependencies. If you really need panic level, use <code>DPanic</code>.<li><p>Use <code>contextID</code> or <code>traceID</code> in every log lines for easily debug.</ul><h3>Don't overuse ORM libs, no need to handle another layer above SQL.</h3><p>Each ORM libs has each different syntax.<br>To learn and use those libs correctly is time consuming.<br>So just stick to plain SQL.<br>It is easier to debug when something is wrong.<p>But <code>database/sql</code> has its own limit.<br>For example, it is hard to get primary key after insert/update.<br>So may be you want to use ORM for those cases.<h3>If you want test, just use <a href=https://github.com/stretchr/testify>stretchr/testify</a>.</h3><p>It is easy to write a suite test, thanks to testify.<br>Also, for mocking, there are many options out there.<br>Pick 1 then sleep peacefully.<h3>Replace <code>go fmt</code>, <code>goimports</code> with <a href=https://github.com/mvdan/gofumpt>mvdan/gofumpt</a>.</h3><p><code>gofumpt</code> provides more rules when format Go codes.<h3>Use <a href=https://github.com/golangci/golangci-lint>golangci/golangci-lint</a>.</h3><p>No need to say more.<br>Lint or get the f out!<h2>Thanks</h2><ul><li><a href=https://github.com/uber-go/guide/blob/master/style.md>Uber Go Style Guide</a><li><a href=https://dave.cheney.net/2014/10/17/functional-options-for-friendly-apis>Functional options for friendly APIs</a></ul>

View File

@ -60,6 +60,51 @@ So in the future, service D which needs to call C will not need to copy libs to
Another bad practice is adapter service.
No need to write a new service if what we need is just common pkg libs.
## Taste on style guide
### Use functional options, but don't overuse it!
For simple struct with 1 or 2 fields, no need to use functional options.
[Example](https://go.dev/play/p/0XnOLiHuoz3):
```go
func main() {
s := NewS(WithA(1), WithB("b"))
fmt.Printf("%+v\n", s)
}
type S struct {
fieldA int
fieldB string
}
type OptionS func(s *S)
func WithA(a int) OptionS {
return func(s *S) {
s.fieldA = a
}
}
func WithB(b string) OptionS {
return func(s *S) {
s.fieldB = b
}
}
func NewS(opts ...OptionS) *S {
s := &S{}
for _, opt := range opts {
opt(s)
}
return s
}
```
In above example, I construct `s` with `WithA` and `WithB` option.
No need to pass direct field inside `s`.
## External libs
### Don't use cli libs ([spf13/cobra](https://github.com/spf13/cobra), [urfave/cli](https://github.com/urfave/cli)) just for Go service
@ -121,3 +166,8 @@ Pick 1 then sleep peacefully.
No need to say more.
Lint or get the f out!
## Thanks
- [Uber Go Style Guide](https://github.com/uber-go/guide/blob/master/style.md)
- [Functional options for friendly APIs](https://dave.cheney.net/2014/10/17/functional-options-for-friendly-apis)