2007.07.15 04:37 "Re: [Tiff] [ANNOUNCE]: Libtiff 4.0.0alpha released", by Ron

On Sat, Jul 14, 2007 at 07:27:01PM -0500, Bob Friesenhahn wrote:

On the JBIG issue, I think it's pretty safe to say that the debian version of libtiff will not include JBIG support. I wonder whether there would be a way to make JBIG support available at runtime on platforms that support dynamic loading. In any case, I think the

A number of years ago I engaged in a discussion with Richard Stallman and the JBIG-KIT author to discuss if using dynamic loading to use JBIG-KIT was acceptable (in order to avoid GPL by providing separability) and the opinion of RMS and the JBIG-KIT author was a solid 'no'.

Yes, I've had similar (though less authoritative) advice regarding mixing GPL software with openssl, which otherwise needs an explicit exception to the GPL for individual works which intend to permit that.

This does seem logical, since if you are using symbols from the library then you are linking to it. That they may be provided statically, or through the ELF dynamic linker, or by explicit dynamic loading from the application (etc.) doesn't really change that much.

So the value of a runtime linkable version would not be related to the software licence. Any lib built with that capability would still be affected by the most restrictive licence of its components (in this case GPL from JBIG-KIT) -- but what it _might_ allow, is distribution of the patent encumbered portions only in places where the distributor is not encumbered by them. So everyone could use the same GPL libtiff, but people in the Free World could also add a separate JBIG component that would be available if present, without needing to rebuild from source.

However, it seems to me that it does allow users who are using GPL compliant distribution methods to extend the package later or remove the component.

The GPL taint by itself is not a problem for Debian per se, though if making the patented parts a runtime selectable option is feasible we may still want to distribute two builds from the same source. A tiff-jbig binary only suitable for GPL apps, and a tiff(-vanilla) package where the existing libtiff licence would be unchanged. If the patented chunks have to be distributed separately, the tiff-jbig enabled package can just have the (GPL symbol) stubs to connect it, with an extra package off site (there are a number of developers with personal sites who may host this, even if there is no non-us option available anymore).

We're surely going to disrupt one or more existing applications if the libtiff licenced option is not available.

There is yet another issue that JBIG-KIT itself is not distributable under the GPL (even though it uses the GPL license) since it violates patents recognized in many countries.

I think the GPL itself tries to be pretty clear on that point. If some external factor means you cannot meet its terms then you have no licence to distribute or use that software. But that does not affect in any way the rights of people who are not subject to those additional factors.

Personally I think if that gives a real competitive advantage to people in countries that don't let patent lawyers and patent collectors plant roadside bombs on the public highway of software development, then more power to them for that wisdom ;-) People who do suffer such dangerous hazards in their daily work should yank on the chain of their legislators if they find that intractable or backward.

If anyone suddenly called in all the patents that exist on anything, then we'd find we have a lot less of just about everything we currently have. The system only "works" because its mostly made up of inert booby traps, just waiting for some individual to become lucrative or threatening enough to trigger the lurking lawyers. So in a system where only people too poor for the trap to spring can actually do anything unimpeded, the natural advantage is going to go to the things they implement.

The particular patent holders identified for JBIG would seem to be of the variety smart enough to realise that a good commodity implementation is better for their overall business than a few piddling remittances from middle size companies. If people in the Free World implement it and get it adopted, they'll know their cue when they see it. In the meantime, the rest of us will just have to wait...

or surf through the tidal wave of lobbyists for the attention of our lawmakers. Sorry for the rant, but I do think we should make it as easy as possible for the people who can use this, to use it. Even if the patent holders do have no sense, their claims won't last forever so if the technology is enduringly good its time will come, if not someone else will come up with something better, and maybe also make it Free from the outset. We'll see. As you say, the race is on ; )

Cheers (from the sidelines),

Ron