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

Re: What will improve Debian most?



On Sun, Mar 29, 2009 at 04:04:22PM +0100, Stephen Gran wrote:
> You're also making some implicit assumptions about what is available -
> are there really 9855 new projects that should have been added to Debian
> last year that weren't?  

Via twitter [0] here's another point of comparison: the iPhone app store
opened in July last year, and in that pretty brief time has something like
30,000 apps [1] for its single platform. Debian has 4,847 less than that
in sid (main, contrib and non-free) for our most popular platform (i386).

CPAN has 15,496 modules apparently, compared to around 1,673 lib*-perl
packages. "gem list -r" lists about 4,500 Ruby Gems, compared to about
530 packages with ruby in the name. There are 4,245 wordpress plugins and
700 wordpress themes. There are about 2,120 drupal modules compatible w/
version 6 (somewhat more for version 5).

Those (CPAN, Ruby, WordPress, Drupal) alone add up to about 25,000
additional packages, let alone 9,855. I don't know how you want to count
the iPhone stuff -- all of which was either ported or written for scratch
in the past ten or so months -- but I suspect most of the software there
isn't already packaged for Debian.

And yes, a lot of those packages would be crap compared to what we
currently have in the archive -- they'd be things that make your phone
sound like bodily functions, NIH-versions of things other people have
already done, Bobby's first program with a deliberately or accidentally
non-free license, and I'm sure there wouldn't be much of interest in
the maintainer scripts for them either.

But take all that for granted: would finding a way to redistribute as
much of that as we could serve our users? I suspect so -- making it easy
to get software, and having a single point of control for sysadmins to
manage it is Debian's raison d'etre, afaict. Would it serve the free
software community? How many modules, plugins and themes are freely
licensed, and how many more would be if they were redistributed by
folks who have learnt about licensing issues and care about encouraging
free licenses?  Wouldn't having a free Linux distro that's keeping up
with new distribution models like the iPhone App Store and Ruby Gems
be a useful tool in keeping free software and Debian users relevant and
competitive on every playing field?

Don't get me wrong, I don't think it'd be easy to handle that sort
of scale -- the archive software might cope with that sort of growth
but it'd be rough, it'd probably be pretty tough on the capacity of
our servers, it'd require more automated maintenance techniques (eg,
automatically packaging every module in CPAN), which would in turn
require more automated testing techniques (so that automatically packaged
modules that are obviously broken don't go into sid), unless most of it
was arch:all stuff, it'd be tough on the autobuilders, it'd be tough on
the mirrors, it'd be especially tough on searching through the Packages
file, and I'm sure there's other challenges that would have to be met.

And maybe Debian's not up to meeting those challenges -- maybe keeping
a roughly constant growth rate is good enough; maybe all the clever tech
that would have to be built is too much effort; maybe there'd just be too
many arguments for anyone to bother with. But if Debian doesn't tackle
the issues blocking us from distributing lots of the software people
are actually using, and if Debian leaves those problems to Apple and
Google and Drupal and CPAN to address... Well, if you're not addressing
the problems people see as relevant, aren't you irrelevant?

Cheers,
aj (wondering if Debian's prospective leaders for the next year will have
    anything further to add when the votion period ends in 30 hours or so)

[0] http://twitter.com/diveintomark/status/1360639404
    http://identi.ca/notice/2904469 

[1] http://www.148apps.com/news/wowza-30000-apps-itunes-app-store/
    http://148apps.com/10000/

Attachment: signature.asc
Description: Digital signature


Reply to: