Re: Bacula and OpenSSL (original) (raw)




On Monday 16 July 2007 10:57, Shane M. Coughlan wrote:

Hi Kern

Kern Sibbald wrote:

On Saturday 14 July 2007 11:03, Simon Josefsson wrote:

That is incorrect. The FSF has granted OpenSSL license exceptions to some software that links to OpenSSL. For example, GNU wget.

Interesting. Shane would you comment on this?

As Simon said, in certain special circumstances selected GNU tools related to networking have been granted exceptions. I suggest that such a grant is not likely in the case of the general purpose code included with Bacula.

However, please bear in mind that I don't speak for FSF :)

Yes, and in addition, after Josselin's email, I did a bit of research, and for at least one of the files that we use (fnmatch.c), the FSF license was changed from GPL to LGPL sometime in 2004 the best I can tell.
Unfortunately, we are still using the old GPL'ed version rather than the newer LGPL'ed version.

While it would be best to convert to the newer version, it is not so easy as the we have made some modifications to the old one to support Win32 systems (stupid difference of path separators, ...).

Question: for old GPL'ed versions of fnmatch.c and fnmatch.h that we are using copyrighted in 1997 by FSF, would it be possible to modify them to use the LGPL as you are currently doing?

That would give me a bit of breathing room (i.e. no recoding for the moment).
Long term, I am probably going to use your newer LGPLed version since it supports UTF-8, but that will take some modifications and lots of testing.

In addition, there are two files copyrighted by Anders Carlsson, two by Sun Microsystems in the tray-monitor directory, and a number of files by AT&T all in the win32 subdirectories.

All of the authors of third-party code would need to be contacted for a linking exception, though from your last email it sounds like you have already removed a substantial amount of the third-party code.

Yes, all the authors would have to be contacted, but given they involve institutions such as FSF and ATT and Sun, I don't plan to go that route, it has little chance of succeeding.

For files like fnmatch.c where FSF has already changed the license, I think it makes sense to ask for a change to the old code. For all the rest, I will work on replacing the files and/or rewriting them myself. It is a terrible waste of time, but it is not a monster project, and the only hard part is the testing that will be necessary to validate the changes ...

I have some availability this week. I could do a physical meeting or a phone call during Wednesday and Thursday afternoon if you think it would be useful. :)

I appreciate your offer, and I think that meeting at some point will be important, especially if FSF is willing to put the older fnmatch.c/h that we use under the LGPL license that is being used by the current version.

However, one important issue to work through is Josselin's claim that due to the wording in GPL v3, I could switch to it, and it would be OK to link OpenSSL in as shared objects.

If that is the case, it would provide a short term solution to this problem so that Debian can continue to release Bacula with OpenSSL enabled in the next version.

Switching to GPL v3 is something I could do before the next release (in a week or two), but I would do it only if I can get FSFE and Debian to confirm that it will resolve the OpenSSL linking problem -- for the moment, I don't have any "official" confirmation from either of you.

Longer term, I am definitely removing all 3rd party copyrighted code that is GPL'ed, and unless GPL v3 turns out to be the magic bullet and does not encomber me with additional constraints, I'll either add back the OpenSSL modification I previously added for Debian, or switch to a less restrictive Open Source License such as Sun's. I'm still looking at other licenses and switching will require a good deal of thought ...

Before swtiching to any other license, should that be the case, and possibly before adding any license modifications, Shane and I will very likely need to meet.

Regards,

Kern


Reply to: