Re: configure not happy with OpenBSD2.3

[Craig Burley <burley at gnu dot org>]

[Jeff Law wrote:]
>Jeffrey A Law wrote:
>> OpenBSD is not supported by egcs.
>> 
>> My understanding of the situation is that the OpenBSD patches do not
>> have a clear assignment path and therefore we can not integrate them.
>> 
>> ie, they were written by someone that for whatever reason did not
>> assign copyright rights to the FSF.  We can not integrate such patches.
>> 
>> Some of the OpenBSD folks have talked about rewriting the OpenBSD config
>> files, filing an assignment with the FSF and donating the code.  But that
>> has not happened yet.

>Obviously I'm missing a significant point here so excuse the naive
>question... Why is it that egcs supports many proprietary platforms that
>provide no source code nor do they assign rights to the FSF but egcs
>does not support OpenBSD for these said reasons even though OpenBSD is
>much more "open" than these proprietary platforms? I'm a Linux/Solaris
>user myself so this doesn't affect me personally but I'm curious about
>your statement.

Jeff was referring to the *patches* that, applied to gcc, make it
support OpenBSD.  Generally, whether the platform for which a given set
of patches enables support is proprietary, or is available in source
code, is orthagonal to whether those patches *themselves* meet the
requirements of the "owners" of the product to which they're applied.

So, it doesn't matter that Solaris or HP-UX, for examples, are
proprietary, as long as the patches (now normal part of the code)
that enable gcc to support them meet the various obligations of
the intellectual-property-related laws and guidelines applied to
gcc.  And if there are no patches meeting those laws and guidelines
that enable support for an "open" platform like OpenBSD, then it
doesn't help much that patches *not* meeting those laws and guidelines
do in fact exist.  (Though individual users can choose to integrate
such patches with their own copies of egcs/gcc, and even redistribute
them -- as long as they meet the requirements of the law.  It's the
FSF that chooses to employ additional guidelines for crucial
products like gcc, and Cygnus that chooses to follow those same
guidelines for egcs, primarily as a preventative measure.)

In practice, "open" systems are easier for products like egcs and
gcc to support than closed ones, and "free" systems (like Linux),
usually, easier still.  And some "closed" systems can be nearly
impossible to legally support, if the information needed to support
them cannot be obtained without signing an NDA that prevents making
use of the information when writing GPL'ed (or similarly licensed)
patches and properly assigning the copyrights for those patches.
(This prevented Linux support for certain video cards from appearing
for a long time.)

In summary: Jeff was talking about the copyright for the *patches*,
not the relative openness or closedness of the operating systems
supported by the patches.

        tq vm, (burley)