代码优化的一点总结
很多时候,为了赶项目进度,或是没有合适的代码review机制,再是每个程序员的水平高低各不相同。以下几点就是最近优化接口时做的一些调整。
- 如果使用循环不停地去操作数据库,请把这个循环放在service层中,并且使用
@Transactional
,这样,就保证多次的数据库操作在一个事务中,不会不断地开启提交事务。不建议要放在controller层或action层中。
举例代码如下:
@Transactional
@Override
public List<Content> collectContents(Set<String> contentIds) {
if (contentIds.isEmpty()) {
return new ArrayList<>();
}
Integer[] ids = contentIds.stream()
.map(id -> Integer.parseInt(id))
.collect(Collectors.toList())
.toArray(new Integer[contentIds.size()]);
return dao.queryContentsInOne(ids);
}
- 如果需要获取redis的多个keys或批量操作时,推荐使用pipeline,减少网络通讯开销。
举例代码如下:
Pipeline pipeline = jedis.pipelined();
List<String> uIds = new LinkedList<>(followingUsers);
for(String userId : uIds) {
pipeline.exists(getFormatKeyStr(USER_CONTENT_ZSET, userId));
}
Pipeline zaddPipeline = jedis.pipelined();
if(Objects.nonNull(followingUpUserContents) && followingUpUserContents.size() > 0) {
followingUpUserContents.forEach(content -> {
zaddPipeline.zadd(getFormatKeyStr(USER_CONTENT_ZSET, content.getUser().getId()),
content.getSortDate().getTime(),
String.valueOf(content.getId()));
});
zaddPipeline.sync();
}
-
熟悉数据表的定义结构,每个字段的类型和长度,以及建立了哪些索引,哪些索引是必需的,哪些是可以去掉的,再是使用explain命令,来查看你要编写的sql语句是否使用了索引。
-
在项目中遇到了OOM溢出,原因是原来Hibernate使用redis作为二级缓存,后因为redis需要大量的网络通信,后更换为Ehcache,结果就OOM了。其原因是Ehcache需要大量内存缓存查询结果,一开始分配的内存并不多。解决办法,增加内存。
-
时刻记得null和为空的情况,可以使用Optional来解决。
举例代码如下:
package crazy;
import java.util.Optional;
class Company {
private String name;
private Optional<Office> office;
public Company(String name, Optional<Office> office) {
this.name = name;
this.office = office;
}
public String getName() {
return name;
}
public Optional<Office> getOffice() {
return office;
}
}
class Office {
private String id;
private Optional<Address> address;
public Office(String id, Optional<Address> address) {
this.id = id;
this.address = address;
}
public String getId() {
return id;
}
public Optional<Address> getAddress() {
return address;
}
}
class Address {
private Optional<String> street;
private Optional<String> city;
public Address(Optional<String> street, Optional<String> city) {
this.street = street;
this.city = city;
}
public Optional<String> getStreet() {
return street;
}
public Optional<String> getCity() {
return city;
}
}
public class OptionalDemo1 {
public static void main(String[] args) {
Optional<Address> address1 = Optional.of(new Address(Optional.ofNullable(null), Optional.of("New York")));
Optional<Office> office1 = Optional.of(new Office("OF1", address1));
Optional<Company> company1 = Optional.of(new Company("Door Never Closed", office1));
// What is the street address of company1?
// In which city company1 is located?
Optional<Office> maybeOffice = company1.flatMap(Company::getOffice);
Optional<Address> maybeAddress = office1.flatMap(Office::getAddress);
Optional<String> maybeStreet = address1.flatMap(Address::getStreet);
maybeStreet.ifPresent(System.out::println);
if (maybeStreet.isPresent()) {
System.out.println(maybeStreet.get());
} else {
System.out.println("Street not found.");
}
// shorter way
String city = company1.flatMap(Company::getOffice)
.flatMap(Office::getAddress)
.flatMap(Address::getStreet)
.orElse("City is not found.");
System.out.println("City: " + city);
// only print if city is not null
company1.flatMap(Company::getOffice)
.flatMap(Office::getAddress)
.flatMap(Address::getCity)
.ifPresent(System.out::println);
}
}
- 如果常见的优化手段并不明显,可以考虑使用临时表。
最重要的一点是:
写代码的时候,不要只盯着代码,要想你写的代码在内存或是在程序中是如何执行的。