[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

transformations of "source code"



On Wed, Mar 05, 2003 at 03:52:20PM -0500, Joe Moore wrote:
> What sort of transformations are permitted?

I'd say any form of lossless encoding that doesn't require a key to
recover, or with which the key is provided.

This definition has a few advantages:
* It's technology-neutral.  cpio vs. tar, gzip vs. bzip, WAVE vs. PCM
  (for audio files), who cares?
* "Lossless" is important; it means you can recover the original data
  stream.
* Encryption is fine (it might be wise, necessary, or even unavoidable
  depending on the distribution channel), but it is the distributor's
  responsibility to ensure that the recipient gets the decryption key.

> There are transformations that should be allowed without making the switch
> from "source" to "object/executable" form.
> 
> For example, source.c.gz (gzip-compressed source file)
> It has been transformed, is clearly derived from the (C) C-source file, is
> not the preferred form for modification, etc.  But is considered
> equivalent to source since most developers can just gunzip it.

Sure; my definition above attempts to preserve that common-sense
understanding.

> Could you imagine an author who said you couldn't distribute
> source.tar.gz, since it's not the "preferred form for modification", but
> rather the "preferred form of distribution of the preferred form for
> modification" (see the difference?)

I'd want to sic Eben Moglen on his ass.  :)

> Well, since the FDL has taken the term "transparent", how about if we call
> the output of tr (or gzip or indent) as "translucent forms".  These are
> derived work that can be mechanically derived from the source form, and
> can be mechanically transformed back to the source form (or near
> equivalent)

I don't think it's necessary to introduce a metaphor, if one can simply
uses my definition for acceptable transformations of the "source form" of
the "Work".

> This does not cover compiling, since even by disassembling the .o file,
> you can't get back the comments, short-term variable names (if any
> variable names), etc.

Right.  Those are lossy transformations -- not lossless.

-- 
G. Branden Robinson                |    I'm sorry if the following sounds
Debian GNU/Linux                   |    combative and excessively personal,
branden@debian.org                 |    but that's my general style.
http://people.debian.org/~branden/ |    -- Ian Jackson

Attachment: pgp8btRu2Ipt5.pgp
Description: PGP signature


Reply to: