snippets [was Re: Respect for Upstream Authors and Snippets of Interest]
> Most non-DFSG-free materials that we find in main are there because
> they were overlooked. I see no reason to suspect the GNU Manifesto
> of being any different.
I think you're wrong about that. Most Debian developers have, I
suspect, read the GNU Manifesto. Its unmodifiable status is not
hidden. Most Debian developers (excepting those unfortunate vi
knuckle-dragger in our midst) know that it can be found down in the
gizzards of the emacs support files. But Debian is full of snippets,
and no one has ever raised them as an issue before. The burden of
proof is really on you here, to show that this was all just some big
inadvertent mistake.
Anyway, I'd like to re-frame this slightly, in order to keep the
discussion focussed.
----------------------------------------------------------------
Okay, I know you understand this, but in order to be clear to others:
*** BY MY DEFINITION:
***
*** A "snippet" is a file in a source tarball which:
***
*** - MERELY ACCOMPANIES and is not an integral part of the source
*** - is REMOVABLE
*** - is NON-FUNCTIONAL (not code, not documentation, not needed for build)
*** - is NON-TECHNICAL in nature
*** - is usually of historic, humorous, or prurient interest
*** - is usually NOT itself MODIFIABLE, eg "may redistribute verbatim"
*** - is very SMALL compared to the technical material it accompanies
***
*** (Good examples of such snippets are historic or humorous emails
*** and usenet posts, political essays, jokes, and the like.)
First, let's divorce this discussion from the GFDL. Separate
question, separate topic.
The GNU Manifesto is such a snippet, in all the {GNU ,x}emacs
packages. But I'd rather use a pretend example in order to clarify
that we're not talking about the GFDL anymore, or RMS or the FSF. So
let's use a copy of the heart-rending email from his cancer-stricken
and now deceased sister that inspired an upstream author to study
molecular biology, work on colon-cancer oncogenes, and write a
biosequence-processing program which is packaged for Debian. It is
typical to find such snippets in upstream tarballs, and to include
them in /usr/share/doc/whatever/. I'm talking about stuff like that.
Stuff that is not modifiable, of interest, reasonable to include, not
code, not documentation, not technical in nature, not part of the
program but merely accompanying it, and small compared to the
technical thing it accompanies. Stuff whose removal would often
impoverish our understanding of the circumstances of a work's
creation, and of the work's author.
If we decide to go on a crusade against them, it would be a really big
deal for a couple reasons:
- Debian is absolutely *rife* with such snippets.
- This is because upstream tarballs are absolutely rife with them.
- It would be a major change in practice.
- Scanning our sources for them would be a gargantuan undertaking.
- No other free software organization eschews such snippets.
- They'd keep sneaking back in.
Furthermore, snippets have not caused any *problem*. We have free
software because non-free software sucks for various reasons. Think
of the RMS printer driver story. We have the DFSG because violations
of them cause actual problems and hassles to actual people, and make
software less useful or modifiable or free. Unremovable invariant
sections could screw up our manuals, and impede various uses and
recycling of material, so we're firmly against them. But snippets in
our packages, in contrast, have not caused anyone any trouble
whatsoever, and by nature cannot! Removing them would be a lot of
work, with no gain. Once we were done there would be nothing we could
point at and say "you can now fix/do/run/help the XXX
program/documentation which you couldn't before." It would not, in
actual fact, increase the utility or freedom of the Debian GNU/Linux
Operating System.
Given all this, it seems highly unlikely to me that we could reach
consensus to change practice and set out to exterminate the snippets.
(On the other hand, this isn't license to sprinkle non-technical crap
all our our beloved distribution. We do have the freedom to remove
the snippets. Like chocolate sprinkles, they should not be overused.
Also like chocolate sprinkles, you don't have to write a formal rule
for when there are too many---it's easy to tell.)
Reply to: