|
Authored by: Anonymous on Friday, May 31 2013 @ 05:55 AM EDT |
"If you just call into whatever part of the code you like, you are not
"following an API". An API is a specification, not something you
figure out by randomly trying things and seeing what works. "
Wrong, the specification is a document, the API is not a specification, the
document is not an API.
If I call into an undocumented memory address and consistently get a useful
result, and then continue to use that effect, than I am using that undocumented
call *AS* an API into whatever chunk of code that I am calling.
This is known art. Cf anything you can find on 'undocumented APIs'
Whether it is official, unofficial, specified, documented is irrelevant to what
I actually do with it.
It becomes an API because that is what how I decide to use it.
Whether my program is stable will remain stable is beside the point.
[ Reply to This | Parent | # ]
|
|
Authored by: Anonymous on Friday, May 31 2013 @ 07:26 AM EDT |
(Christenson) [ Reply to This | Parent | # ]
|
|
Authored by: Anonymous on Friday, May 31 2013 @ 08:46 AM EDT |
1) Your ability to post a comment has nothing to do with API and much to do with
protocol. HTTP POST specifically, or possibly PUT. No API involved.
2) Compaq did not write its own implementation of IBMs API.
Compaq wrote it's own implementation of IBMs BIOS, in the process produced a
level of compatibility in it's interfaces that meant at the friction layer
(where the rubber meets the road) it was essentially the same as IBMs. No API
involved.
[ Reply to This | Parent | # ]
|
|
Authored by: Jaywalk on Friday, May 31 2013 @ 09:47 AM EDT |
Back in the day (I feel older every time I say that) Microsoft had a bunch of
APIs that they never documented. Supposedly, that was because they might change
the APIs and didn't want to promise that they would work the same way in the
next version. It's been alleged that there was a hidden agenda allowing MS
programs to work only on "real" DOS and not on competitors like
DR-DOS. Andrew Schulman made a good living documenting the undocumented and you
could only get some stuff to work by breaking (or at least bending) the rules.
The point is that you can make an interface into a program with or without
documentation. If your program interacts with mine, there has to be an
interface of some kind. Saying that using my interface gives me some kind of
ownership rights over your program makes about as much sense as a car company
saying that using a steering wheel gives them rights over the driver. Reading
the car's manual before you drive doesn't change the equation.
---
===== Murphy's Law is recursive. =====[ Reply to This | Parent | # ]
|
|
Authored by: Anonymous on Friday, May 31 2013 @ 03:16 PM EDT |
But interfaces are not all APIs.
Your living room has an interface from the outside world known as a front door.
The authorized API for this interface specifies a protocol for knocking,
verifying personage, then opening the door.
I can break a window and enter, thereby creating my own interface, but I have
not created a new API.
Oracle's problem here is that they accuse Google of breaking a window and
entering, but their "evidence" consists of a door and a doorknob with
an "out to lunch" sign. Not to mention that so far it looks like
Google merely copied the door, but never actually entered the room...[ Reply to This | Parent | # ]
|
|
|
|
|