Cygwin license

Steve Morris
Tue Mar 16 16:30:00 GMT 1999

DJ Delorie writes:
 > Linux does not live on top of the HURD.

On the Alpha? I had heard that it did. I don't mean the Intel Linux

 > > So if I develop code from scratch in cygwin, compile it in cygwin and
 > > want to distribute the binaries, free, to users of the sourceware
 > > version of Cygwin, Cygnus says no. Cygnus is suggesting that I don't
 > > have the right to give away non GPL binary code to the users of the
 > > cygwin development environment without giving them money for the
 > > privilege. All because I need to compile in the "glue" required to
 > > interoperate with other programs in the cygwin.dll platform.
 > We don't consider cygwin to be "glue".  There is a substantial amount
 > of work and new functionality in cygwin.  Calling it "glue" is
 > insulting to the people who created it.  

Please stop puting words in my mouth. I am trying to stick to
facts. No insult is intended. I do not refer to cygwin as glue and
never have. It is obviously an impressive and high quality
accomplishment. I am extremely glad for its existance. 

However it is my understanding that most of cygwin lives in the
dll. The interface to this dll, i.e. the headers and the dll interface
library is what I am referring to as glue. glue to connect the
executable to the cygwin environment. I am not suggesting that the
entire cygwin package is merely glue to win32. Those are your words,
not mine.

I am suggesting that the interface library and headers are legally
different from the dll itself and may have different legal
requirements concerning licensing. dlls are often distributed
seperately from the interface "glue" to the dll. Most of the dlls in
use are part of Windows. These are not considered part of the
executables that use them. This is a distinction that you are

I contend that cygwin comes in two parts. 1) The cygwin.dll and 2) a
thin interface layer to that dll. Obviously Cygnus has full rights to
control cygwin.dll. It is not so obvious that they have equivalent
rights to control the interface to that dll. Interface code that
provides no "value add" other than a connection to an underlying
package tends to trip on interoperability requirements. The underlying
package can be controlled but the interface usually can't. Wether it
is in this case is a matter of law, possibly untested law.

 > We wrote a piece of software
 > that *adds functionality* to Windows.  If you want to use that
 > functionality, you must abide by our terms.

Only if those terms are legally enforcable. That is the subject under
discussion. Licenses only make sense in a legal context. Just like a
manufacturer cannot disclaim all warrenty for their products, Cygnus
cannot arbitrarily restrict use of their code. Whether this particular
restriction on the interface to the dll is legally valid is the topic
under discussion. It may be. Cygnus clearly believes it to be so. That
doesn't prove it.

 > You're grasping at straws.  The fact that the user has N cygwin
 > programs on their machine already doesn't change the fact that your
 > program uses cygwin and *includes* code from cygwin - code that does
 > more than just "glue" you to the OS (win32).

I am not grasping at straws. Please bring this conversation back to
the level of discussing facts and relevent opinions. I am eithor right
or wrong. In neithor case should you insult me. Counter my arguement
if you think I am wrong but please leave out the personal attacks. I
am merely stating my honestly held understanding.

In some sense I am a disinterested outsider since I have no plans to
or interest in providing non sourceware packages that run under
Cygwin. (I speak as if I did but that is merely a hypothetical.)
However I do not GPL my code. I provide an unlimited source license to
it with the usual disclaimers.

I merely happen to be an outsider who has some level of understanding
of the law. It is interpretation of the law that we are discussing.
Your persistance in misinterpreting my position has strengthened my
desire to make that position more clear, otherwise I would have
dropped out of this conversation and left it to interested parties.

To the extent that I have any interest at all I would like to see
Cygnus make money out of cygwin. I would pay a moderate fee for its
personal use at home. I would recommend to employers that they license
cygwin from Cygnus. In the past I have even provailed on a similar
recomendation for the Cygnus gcc cross tools. However I also believe
that the interoperability issue applies to cygwin.

So in summary:

1) I am not suggesting that Cygnus cannot control distribution of the

2) I don't think that cygwin is a trivial glue layer. My opinion is
quite to the contrary.

3) I wish Cygnus well and hope they make as much money as the law

4) I believe that it is possible that the law would allow the
unlimited distribution of code linked only to the dll interface as
long as the dll itself was not part of the distribution. The end user
then has the obligation to follow the cygwin GPL requirements,
i.e. they must legally acquire its use.

5) I believe that challenging legal positions has as much use and
validity as testing the security of encription or access. The law must
be challenged to be useful. Cygnus needs to stand on a strong legal
basis or it will not stand at all. Similarly software security must be
tested to find weaknesses. In neithor case is ill will
involved. Offense is not intended and shouldn't be taken. You would
not be offended if I suggested that there was an authentication flaw
in cygwin software. You shouldn't be offended if I similarly suggest
that there is a flaw in cygwin licensing. Both are merely theories
which need testing and possible bug fixing.

Want to unsubscribe from this list?
Send a message to

More information about the Cygwin mailing list