Matthew Garrett wrote: > Don Armstrong <don@debian.org> wrote: >>It's not either. It's a hypothetical. That is, if, hypothetically, the >>source provided is the result of a obfuscation regex, then it's not >>source. [IE, we aren't provided the real prefered form for >>modification.] > > While the GPL defines source as the prefered form for modification, that > definition doesn't exist in the DFSG. There's no reason to believe that > we need the preferred form for modification, merely an acceptable form > for modification. Otherwise we run into all sorts of issues with JPEGs > and suchlike... "acceptable form for modification" will get you in even worse trouble than "(author's) preferred form for modification". The former is a subjective criteria, and could raise issues with any code that someone claims is difficult to maintain (due to lack of documentation, poor programming practices, obscure language, any arbitrary criteria you might think of for unmaintainability). The latter is an objective criteria, which will only ever trigger in cases of obfuscation and/or compilation. We do need some ability to determine if we have real source code available; "preferred form for modification" seems like a well-established definition, and far better than the alternatives. I don't think we'd be having this conversation if the code was truly obfuscated, in the sense of deleting all whitespace, using #defines to obscure code structure, changing all control structures to gotos, etc. The only reason we're hesitating is that we aren't 100% sure that the author hasn't just written the code this way because they have the documentation in front of them and it all makes perfect sense to them. I don't think intentionally obfuscated code passes the source code requirement of the DFSG any more than a compiled binary does; if it does, we have a problem. Undocumented code, on the other hand, while rather annoying, is not an issue of freedom. - Josh Triplett
Attachment:
signature.asc
Description: OpenPGP digital signature