https://bz.apache.org/bugzilla/show_bug.cgi?id=59836

            Bug ID: 59836
           Summary: Replace primitives with enums
           Product: POI
           Version: 3.15-dev
          Hardware: PC
                OS: Linux
            Status: NEW
          Severity: enhancement
          Priority: P2
         Component: SS Common
          Assignee: [email protected]
          Reporter: [email protected]

This is a aggregate bug for recent changes in classes from using short or int
to using enums with codes.

Benefits:
* self-explanatory usage rather than needing to list the valid constants in the
javadocs
* easier to distinguish arguments in function signatures (not just a 5 ints)
* clearer error messages, no functions needed to convert code to string
* free range checking (less boiler plate code that does the same)
* type safety

Drawbacks:
* some permgen overhead
* type checking execution overhead

Overall, this seems like a good thing.

In effort to maintain backwards compatibility, the following is suggested:

If there is currently
int getSomething();
void setSomething(int);

Then we should add
enum getSomethingEnum();
void setSomething(Enum);

And deprecate
void setSomething(int);
int getSomething();

It's a bit tricky handling getSomething/getSomethingEnum while maintaining
backwards compatibility because at some point the signature will abruptly
change. Both getters should be offered for a while with a note encouraging
users to use the constants or enums rather than using literal values. This is
good coding practice anyway to not hard-code literal values, but is especially
important with regard to backwards compatibility here.

At least 2 releases later, but not later than POI 4.0:
Delete void setSomething(int);
Delete int getSomething();
Rename enum getSomethingEnum(); to enum getSomething();

In the end, we will have
enum getSomething();
void setSomething(enum);
as the only two methods.

-- 
You are receiving this mail because:
You are the assignee for the bug.

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to