2004.07.08 23:14 "[Tiff] LZW and libtiff", by Bob Friesenhahn

2004.07.10 17:31 "Re: [Tiff] LZW and libtiff", by Jay Berkenbilt

Now that the Unisys LZW patent has expired world-wide (as of yesterday), what is libtiff's position on including LZW support in libtiff? If source for LZW is included by default, will support for LZW be enabled by default in libtiff?

I can merge LZW code back to the main trunk, if nobody can't give a reason not to do that. It is not an issue for me, because there are no software patents in my country, but LZW code was separated to help users to avoid problems in their countries, so if someone has opinions against LZW, please, tell us.

All libtiff codecs and features are enabled by default, and LZW will do after inclusion.

This topic was lightly discussed in the context of GIF on the debian-legal mailing list. See this message and its references and follow-ups:

http://lists.debian.org/debian-legal/2004/07/msg00091.html

If there was a consensus, it seems to be that it's okay to consider the LZW patent over since IBM never enforced their patent. One person also states that the patent is actually for a related but different algorithm (LZMW) which, like LZW, is an LZ78 derivative.

The site burnallgifs.org also now starts with, "GIFs are now patent-free, but you can learn from the GIF fiasco and protect the digital information you create from patent attacks."

Apparently, the GD Graphics Library is going to be reactivating GIF write support shortly. (http://www.boutell.com/gd/)

ImageMagick also appears to be re-enabling LZW by default (http://studio.imagemagick.org/magick/viewtopic.php?p=192) though it's hard to say for sure since this announcement is old.

Most of the discussions I've seen about this topic are about GIF. People seem to forget about LZW compression in TIFF. This is important as it is the best non-lossy compression algorithm supported by virtually all TIFF readers. (Deflate is not as widely supported, JPEG is lossy.) But of course everyone here knows that already.

The League for programming freedom also seems to think that the LZW patent has really expired (http://lpf.ai.mit.edu/Patents/Gif/Gif.html) and doesn't mention the IBM patent at all.

I'm not a lawyer, but there does seem to be a preponderance of evidence that it would be safe to consider the LZW patent issue over and include it by default in libtiff once again.

If this happens, what is the schedule for the first release that will include it? Will there be a new 3.6 release with it, or will we wait for 3.7.0 for a non-alpha release that includes LZW by default? (I know, that's not really a fair question.)

Jay Berkenbilt <ejb@ql.org>
http://www.ql.org/q/