|
Authored by: wharris on Monday, June 03 2013 @ 11:14 PM EDT |
I guess with all the trolls around, I should clarify that an API can be used for
some purposes without an implementation, but a working program requires an
implementation. I'll try to be more careful with my wording. "Use" is
such a generic word it's easy to misinterpret.
[ Reply to This | Parent | # ]
|
|
Authored by: PolR on Tuesday, June 04 2013 @ 12:07 AM EDT |
Two things.
1- The proper technical term for this implementation is not SDK. Even though it
may sometimes be included as part of an SDK we still should not use the term SDK
to refer to the implementation because there are ways to implement the functions
of APIs without making a SDK. In many cases a simple library file would do.
2- It is true that without the implementation the API is useless. But the point
of the implementation is not to make the API useful. It is the other way round.
We need the API to use the implementation. If we don't convey this message, what
stops a judge from ruling there is no problem copyrighting APIs because code is
needed to use the APIs and not the other way round? He may say programmers can
just arrange their code to use another API because changing the API is not
blocking them from using their code.
The answer is that the API is a choke point. If we write an alternative
implementation, we still need the exact same API to use the new code. Hence, we
need to say the API is required to use the implementation. When we put things
the other way round we don't explain what is the choke point.
So when PJ says we need the SDK to use an API she is wrong on both points 1 and
2.
[ Reply to This | Parent | # ]
|
|
|
|
|