On Monday 05 November 2007 08:17:50 Michael Gerdau wrote:
> [Licence arguments from Paul Millar skipped]
> Not sure whether that's valid or not, the main problem is, who would
> defend such a likely controversial position in court ?
> Possibly against a Billion-$ company ?
> [note this is an old argument raised several times in the past; note
> also the keyphrase "likely controversial"]
Well, to me it seems you have painted yourself into a corner here.
Ultimately, all information about the Win32 API must come from Microsoft; t=
developed the API, after all. Therefore, all information contains in the=20
w32api package must be derived from some Microsoft source or other.=20
If, as you say, the factual details of Win32-API (GUIDs, etc) attract=20
copyright, one could ask where is the license from Microsoft authorising=20
MinGW project to distributing any of the API within win32-api?
I believe this "public domain" business is an irrelevance: a distraction th=
is holding back MinGW. All pure factual information is public domain. Onc=
committed to a medium (as a book or computer file) or stored in sufficient=
great a number (e.g. a huge database of telephone numbers) it attracts=20
copyright, but no one can copyright small isolated facts: the (empty) sky=20
(during day-time) is blue, grass is green, IID_IBindHost GUID is 0xfc4801a1=
0x2ba9 0x11cf ...
But, to answer your question, who would defend [this] likely controversial=
position in court? Well, currently the MinGW developers do, although they=
may not realise it. If the win32-api package contains information taken fr=
MSDN, it has information taken from a copyrighted source. I believe all MS=
web pages even have a copyright statement (at the bottom of the page); the=
one I just looked at had "=C2=A9 2007 Microsoft Corporation. All rights res=
> > I honestly believe it to be acceptable to take patches based on other
> > headers considering the cases that have occured in the US. But if you
> > do not believe so, what would be an adequate solution?
> I'm again iterating an old argument:
> Iff the information in question had been obtained in an illegal way,
> then having it made publily available (even under a licence compatible
> with MinGW, which in itself is questionable AFAIAC) does by no means
> "suddenly" constitute a situation where this information may now be used
> freely by interested parties.
To me, this sounds like you are confusing the argument by misrepresenting=20
what's being said, i.e. setting up a "straw man" :
Nobody is suggesting that information, if obtained illegally, suddenly beco=
legal once distributed widely.
Also, I'm sure you don't mean to question the legality of code within the =
Wine project or suggest that the information might have been obtained=20
illegally. They take legal advice from the Software Freedom Law Center, wh=
have conducted an exhaustive review of their code-base.
But, back to the main point: the argument is that pure facts are simply not=
copyrightable. Microsoft may grant one a license that allows (or prohibits=
copying of their software, files, data, etc. However, they cannot prevent=
someone from repeating any purely factual information contained in their SD=
> To illustrate this by an example:
> Assuming you could somehow get the "secret" formula of how Pepsi or Coca
> Cola are made, you publish them somewhere and put them in the public doma=
> would probably not entitle a 3rd party to start copying it and get away
> with it (at least not in the part of world acknowledging trademarks and
I'm afraid this is a vacuous example and has nothing to do with the w32 API.
Any such "secret formula" would be protected under the country's trade-secr=
act(s); these are a completely different set of laws from copyright. Thos=
governing trademarks are another set of laws, patents are yet a third set. =
One cannot group them together when discussing copyright.
> As I see it:
> Wine and MinGW could and IMO should share patches iff Wine could show
> us (convincingly ;) they retrieved the information by other means than
> looking into M$ PSDK.
What other source would you accept? Can you be sure whatever source you=20
accept didn't get their information from Microsoft SDK or some other=20
Back to the point: this is all irrelevant: simple facts are not copyrightab=
No copyright has been broken as none exists.
> AFAIAC even getting the values by means of a debugger/reverse engineering
> would be fine.
Well, you may think so and you may be right for pure factual information. =
This is how clean-room implementations "work" (i.e. are legal).
However, I believe many (most?) inter-opt projects never accept code from=20
people who have reverse-engineered the original code: the risk of=20
contamination with copyright material is too great. Certainly, a developer=
who has reverse-engineered Windows or viewed Windows source code is barred=
from contributing code to Wine.
> > I'm really hoping we can find some common ground. Mingw is of great
> > use to wine so it'd be great to both give something back to mingw and
> > also make it easier to use mingw for testing wine.
> I'm in perfect agreement with you on this. My only problem is that I
> suspect some of Wine's headers had been created by looking into the
> PSDK headers. And that is out of bounds as far as MinGW is concerned.
I, too, hope this can be resolved: I would really like to see MinGW's supp=
for Win32 improve.