我应该使用os.Open使用log.Panic（）还是log.Fatal（）吗？

问题描述：

when we have:

f, err := os.Open("no-file.txt")
if err != nil {
    log.Panic(err)
}
defer f.Close()

I think it would make more sense to use log.Panic(err). Right? Panic() allows defer f.Close() to execute but log.Fatal() prevents it.

Or the file doesn't get opened if not found? I supposed in that case it's irrelevant if we use Fatal or Panic. Right?

当我们拥有以下内容时： p>

  f，err：= os  .Open（“ no-file.txt”）
if err！= nil {
 log.Panic（err）
} 
defer f.Close（）
  code>  pre> 
 
  我认为使用log.Panic（err）更有意义。 是吗？
Panic（）允许延迟f.Close（）执行，但log.Fatal（）可以阻止它。 p> 
 
 
或者如果找不到该文件则无法打开？ 我认为在那种情况下，如果我们使用致命或恐慌是无关紧要的。 是吗？ p> 
  div>

答

log.Fatal() should rarely be used in production apps–if ever–as it terminates the whole app. log.Panic() panics after performing the log, which again is rarely desired.

Many examples use them (or alternatively a single panic(err) call) to make the example code shorter (to let you focus on what the example is about), but they should be used sparingly in production apps. Instead handle the error "properly". What that means is use case-specific, you may choose to log it and return, or return a new error or do something else, but only do one thing with it (only handle it once). See Writing good Golang code.

答

I prefer log.Panic().

The log.Panic vs log.Fatal is essentially panic vs os.Exit(1).

Exit is worse that Panic.It also makes testing much harder.It’s much harder to deal with a code that does os.Exit.It’s quite simple to stop panic in test by the means of recover.

So, It's better to choose something that does less potential damage.

我应该使用os.Open使用log.Panic（）还是log.Fatal（）吗？

相关推荐