Bug#261257: ITP: folding -- Folding@home Client (install
sean finney wrote:
hi nick,
On Mon, Jul 26, 2004 at 06:37:49PM -0400, Nick Lewycky wrote:
>This packages is so non free that cannot even be included in the
>non-free section without an installer. It's also only for i386.
>dpkg will not be able to track its files.
"without an installer"? This package *is* an installer!
i believe he means, "you can not even re-distribute the binaries, you
must
distribute an installer", similar to what users have to do for the
non-free flash player or nvidia's non-free hardware acceleration
binaries.
Then we all agree.
And what's this about dpkg not being able to track its files? Was that
statement intended to be disconnected from the i386 specificity?
i think you weren't paying close enough attention to what he was
implying. if the .deb package contains only a shell script that wget's
the binaries from the stanford site, dpkg has no way of tracking what
files are part of the software, and there isn't any way around that that
isn't a kludge or really complicated. the nvidia package is a good
example of the Right Way to do this (downloading stuff, building it into
a new package, and letting the admin install that), but that counts
as "really complicated", and certainly not something i would consider
a good idea for a first time package maintainer.
dpkg doesn't, but that's mostly irrelevant. Debian Policy provides me
with enough detail to correctly track the files manually, by the postrm
and postinst scripts. My package does so.
I consider the nvidia package to be Broken As Designed, and when I had
NVidia hardware, I avoided it simply because of that. Is there a good
explanation of why users should be required to have a complete debian
package build toolchain just to use a package? I am capable of changing
my mind.
Another one of Debian's essential interests is a commitment to its
users. Folding@Home has a community of around 300,000 users and is
growing. It was only a matter of time before the two groups intersect.
I
want a Debian package for it. My choice was to either RFP it or ITP it
and I chose the latter. If I'm capable of packaging it, why wouldn't I
choose to contribute?
every now and then someone brings up on this list that a piece of
non-free software should be essential to debian because it makes it more
useful for its users, and they tie this is with the first half of clause
4. i don't understand how you can get down to #4 without reading #1, or
even the second half of #4 though...
Relax, I've read the whole thing. Please don't imply that I haven't.
I never said that the software was essential. Far from it, I explicitly
state that it's an optional, miscellaneous contribution. And I won't let
you turn this into the damned "remove non-free from Debian" debate. Even
Brandon needs a break.
if you're really interested in packaging this particular non-free
software, please consider either a) convincing the authors to make their
software dfsg free, or b) providing a real debian package that the
authors can offer for download from their site or apt-repository.
i think that in general the "wget-install" debian packages are something
to be avoided, and you'd do a better service to both communities if you
could make a better package for those who wanted it.
a) No, it'd be a disservice to the Folding community. Their FAQ explains
that they need to keep the code secret for scientific integrity:
http://www.stanford.edu/group/pandegroup/folding/faq.html#project.source
If it bothers you so much, then the molecular dynamics packages that
Folding@home uses for its simulations are both available under DFSG-free
licenses. But those aren't what I'm packaging.
b) Yes, I plan to try that only if Debian refuses to accept the package,
which could happen if no one sponsors it. Honestly, I don't expect the
Pandegroup (Folding@home upstream) to respond to me at all.
Nick Lewycky
Reply to: