On Thu, Jun 13, 2002 at 12:16:56PM -0500, Branden Robinson wrote: > On Wed, Jun 12, 2002 at 09:56:13PM -0700, Ben Pfaff wrote: > > If the original image is an XCF that takes advantage of features, > > like layers, that PNG does not support, then sure I'd ideally > > want the original XCF if I was going to modify the icons. > > Similarly, some documents have features that get lost in > > translation from one format to another. If the original is in > > Texinfo and I only get a version in HTML, then there's going to > > be information lost. Less drastically, I'm sure there are > > features in each of, say, DocBook, Texinfo, and LaTeX that are > > not present in the other two formats, so that for modification > > the original format is preferred.
> > I like the way the GPL defines source: "preferred form of the > > work for making modifications to it". > I sympathize with Joe's position to some extent, but I agree with Colin > and Ben. Moreoever, it's far too easy to abuse this latitude. Just > because a format is "open" doesn't mean it can't be obfuscated. > Take, for example, the binary-only firmware in some Linux kernel > modules. I don't know how they're licensed, exactly, but even if > they're licensed under MIT/X11 terms that's not good enough for > practical purposes. Machine code may be an "open" format but it's too > hard to work with. Assembly language isn't, if it was written in that > form, especially since assembly tends to be very heavily documented > (otherwise, the programmer gets lost). OTOH, "preferred form of the work for making modifications to it" is not always the same thing as "original source form of the work", because *preferences* are subjective. I may receive a copylefted document in LaTeX format, but because I'm not comfortable with working in that format, I use one of the many available tools to convert it to XML (or God forbid, PDF) before making my own changes to it. Should I then be obliged to redistribute the original LaTex document that I received under a copyleft license? Analogy in the coding world: if there's a particular function (C file, library, etc. -- whatever the basic unit is that we'd like to consider) that's being poorly optimized by the compiler, and I reimplement it in assembly because I know better, should I be required to distribute the original C source, or can I toss it? Personally, I think anyone who takes a piece of portable code and modifies it to be architecture-dependent is a jerk, but I'm not sure that opinion should have the force of law. Steve Langasek postmodern programmer
pgpA0McC7qVD8.pgp
Description: PGP signature
