Skip to content

Latest commit

 

History

History

error_code

Folders and files

NameName
Last commit message
Last commit date

parent directory

..
flightservice
flightservice
 
 
README.md
README.md
 
 

README.md

std::error_code

Understanding standard library error code and try its usage, i.e. std::error_code, std::error_category, std::error_condition.

Sample code

It's a flight service comes from Your own error code, see my implementation in flightservice.

  • requirements for the flight service

    • It's used for looking for flight connections. You tell me where from and where to you want to go, and I will offer you concrete flights, and a price.
    • In order to be able to do this, the flight service calls other services in turn:
      • one for finding the (short) sequence of flights that will take you to your destination,
      • one for checking if there is still seats available on these flights in the requested class of service (economy class, business class),
    • Each of these services can fail for a number of reasons.
    • See details in Your own error code.

  • coding style

$ cd flightservice 
$ clang-format -style=webkit -i *
  • run test
$ cd flightservice 
$ g++ -std=c++14 main.cpp flights_err.cpp seats_err.cpp failure_source.cpp && ./a.out
FlightsErr:10
  value: 10 msg: nonexistent airport name in request
  FailureSource message: invalid user request
SeatsErr:20
  value: 20 msg: all seats booked
  FailureSource message: no solution found for specified request

Note

Convention

  • always use 0 to indicate OK/Success

Understand types

  • std::error_code
    • used to store and transport error codes
    • it has a int to store error code it self, and const std::error_category* to store the error_category of this error code
    • a error_code will be either print to log or compare with error_condition
  • std::error_category
    • used to store which subsystem the error comes from, and give concrete message of each code value of the subsystem
    • when call error_code.message(), the concrete message will be returned due to its polymorphic properties
  • std::error_condition
    • used to match errors
    • callers will only handle errors by check whether error_code belongs to error_condition(the meaning is 'belongs to' though using ==)

How to use std::error_code/error_condition in your own application

See example in flightservice.

  • define your own error code/condition enum, e.g. enum class xxx
  • specialize the std::is_error_code_enum/is_error_condition_enum template for your error code/condition enum
  • overload the make_error_code()/make_error_condition() function for your error code/condition enum
  • implement your error category(inherit from std::error_category) for your error code/condition enum(so that it's possible to print subsystem name and concrete message into log)

Pros and Cons

  • Pros

    • able to store error code itself in a int
    • able to distinguish which subsystem the error comes from
    • able to print concrete code value and message to log
    • callers only need to handle errors they care (std::error_condition)
      • in the flightservice example, callers only want to distinguish whether error caused by BadUserInput/InternalError/NoSolution instead of handle many FlightsErr/SeatsErr.
    • able to use ==/!= to compare std::error_code and enum xxx directly
    • able to construct std::error_code by enum values (e.g. std::error_code ec = XXXEnumType::Value1)
    • std::error_code only contains two members(a int code and a const std::category*), lightweight

  • Cons

    • difficult to understand
      • i.e. difficult to understand the whole story that why design it like this(maybe due to too complex requirements of standard library)
    • difficult to use
      • e.g. library/application developers have to do a lot of work to plugin new defined enum to std::error_code
      • e.g. users have to understand the std::error_code design before use it
    • values of std::error_code and std::error_condition are not the same
      • generally when we do a == b, in intuitive a and b have same type and same value, but they're not.
      • it'll very confuse if we print both std::error_code and std::error_condition into log.

Ideas for design good error code

std::error_code is designed for C++ standard library, which may have a lot of complex requirments to make it so complex like this.
When we need a error code design for our own library/application, there could be some ideas can help us to get a simpler but sufficient design.

  • follow the convention that 0 always means OK/Success
  • use int to store error code
  • each error code has a corrsponding string message for better logging
  • make error code consistent(a int code only has one meaning)
  • distinguish subsystem by use different code segment since we possible to pre-define the segments(e.g. 1000~1999 indicate A subsystem, 2000~2999 indicate B subsystem, etc.)
  • simplify error handling for caller

Remain questions

  • you can teach a new error condition to recognize existing error codes, but also you can teach a new error code to be recognized by existing error conditions
    It's metioned in Your own error condition. Not quite understand it.

References