You are both mistaken, sorry.

You are both mistaken, sorry.

Posted Aug 19, 2004 18:17 UTC (Thu) by mmarsh (subscriber, #17029)
In reply to: You are both mistaken, sorry. by hummassa
Parent article: IBM's summary judgment motion

My understanding of the GPL is that if you write a piece of code from scratch, it's yours to license as you see fit. If you link it against a GPL'ed library and distribute the resulting binary, then you must abide by the GPL for that work, necessitating the release of your source code. If you release just the source, and no GPL'ed code is included, then you can use any license you like, regardless of how the users compile and link the code. Linking thus only applies to binary distribution.

Returning to the subject of IBM's claim re: AIX and SVRX, even under the GPL, IBM's code for AIX would not be a derivative work of SVRX, even though the OS as a whole, when compiled and linked, is. Using the example of a program that calls printf, writing code that calls printf doesn't create a derivative work, since it could be linked against any library that provides a function called "printf" that has the same semantics. Only when distributed with the actual _version_ of printf from glibc does the latter's license come into play. At worst, IBM might have to accept some relatively small amount of "glue" code in AIX as derivative of SVRX, if SCO could convincingly argue that the semantics of the call to SVRX code are sufficiently unique. Even then, we'd be talking about individual lines, and possibly some structs, that would need to be sanitized to become free of the SVRX restrictions.

---

(Log in to post comments)