Repository navigation
Getter is setting the value to null if the key does not exist #754
Description
Activity
Closing this since the documentation already recommends at() vs. operator[], although this definitely is unexpected behavior.
In function from_json, use function at() to access the object values rather than operator[]. In case a key does not exists, at throws an exception that you can handle, whereas operator[] exhibits undefined behavior.
Note this behavior is exactly the one of
std::map:#include <map> #include <iostream> int main(int argc, char const *argv[]) { std::map<std::string, int> m; m["xxx"] = 1; auto y = m["yyy"]; for (auto i = m.begin(); i != m.end(); ++i) { std::cout << i->first << " : " << i->second << std::endl; } }
produces:
xxx : 1 yyy : 0The note you cited refers to the case when we access a const object. Then,
operator[]cannot addnullvalues to the object when called with an unknown key and the semantics is undefined.Thanks for your comment. I wonder why std::map behaves this way. It is counterintuitive to set the value to null or 0 when the key is missing.
Because if it didn't, then
json["xxx"] = 1;wouldn't work. It creates an entry with the default value, and then sets it to1.Hi,
First of all thx for this library it is great.I am puzzled by the answer given to this issue.
In the readme there are a ton of sexy examples using a json variable as a map. They look indeed great, but then you are saying here "if you follow the sexy syntax we claim to have in the readme then you will run into undefined behaviour when one day you'll ask for a non-existent key. You should not follow the sexy syntax we show and access elements with an .at() function" ???
Either the readme should show .at() examples or [ ] should be a little bit safer no ? Am I missing sth ?
There is already a note, see https://github.com/nlohmann/json/blob/develop/README.md#notes.
Here is the sample code
The output is
{"xxx":1,"yyy":null}.The expected behavior is to throw std::out_of_range exception