On 08 Aug 2002 14:59:43 -0700, [EMAIL PROTECTED] (Thomas Bushnell, BSG) wrote: >I like the DFSG-4 fine. Indeed, it covers TeX, and crucially so, >because tex.web permits modification only in the form of patches. >Similarly, cmr10.mf permits modification only if you change its >filename. Both of these are relatively harmless, in large part >because they don't affect what we name commands or filenames in the >installed system. Accordingly, DFSG-4 is a great thing! It correctly >labels TeX (and, I see now, CM fonts) as free. > >But if we cannot install what we want on the system, when it comes to >the functional elements of the programs (not what is displayed in >banner messages, for example) then the software simply is not free. >TeX actually *does* allow us to do this. > >And the LAPL just doesn't. The LAPL is, in fact, *different*. Not by >some mysterious thing, but it's manifestly different. We can quote >the terms!
I have to agree with you on the licenses for TeX and MF (the programs); their modification conditions are quite different from those of the LPPL (but they are mostly stricter), and the first two sentences of DSFG:4 clearly allow the restriction that tex.web and mf.web must not be modified (even if renamed first). OTOH if, as I recall it, Branden should get his way and (at least) these two sentences were to be removed then TeX and MF would be hopelessly non-DFSG-free. I also somewhat have to agree with you concerning the integrity of a TeX system; the copyright licenses on TeX and MF cannot require that plain.tex and plain.mf are the files Knuth made. I can imagine that the trademark licenses can require something of that sort, but IANAL. Peer pressure is probably more efficient too. However concerning the CM fonts I think you're wrong, since the conditions for these are indeed very similar to those of the LPPL; it's just the case that the LPPL relaxes these conditions in some cases. If you think a "rename file before modification" rule is clearly DFSG-free then I'm glad, but there are others that need to be convinced of that too. Observe that cmr10.mf inputs cmbase.mf and roman.mf. roman.mf inputs romanu.mf, romanl.mf, greeku.mf, romand.mf, romanp.mf, romspl.mf, romspu.mf, punct.mf, accent.mf, and at least one of romlig.mf, comlig.mf, and romsub.mf. The code that makes the M is in romanu.mf. Hence if you want to change the M then you have to make a copy of that file under a different name and modify that. You also have to do the same with roman.mf and cmr10.mf, since the modified romanu.mf will otherwise not be used. Some people on this list expressed the opinion that if you can do X by modifying one file if you don't change the name then it is unreasonable to require a name change if a name change means you have to change two or three files to do X. I don't agree with that, but it was nontheless (at least at some point) a central objection to the LPPL rename condition. Lars Hellström
