Re: [ANNOUNCE] debpartial-mirror Alioth project

[dann frazier <dannf@debian.org>]

On Wed, Jun 30, 2004 at 04:58:01PM -0300, Otavio Salvador wrote:
> 
> This can resolved deps, virtual-packages, have regexp filter support,
> can get files, whole directories, ... so on ;-)

Having obviously not done due-diligence to look for a similar project,
I wrote something for a similar purpose as a wrapper around debmirror
a couple weeks ago - here's a brief description.

I can post bits somewhere if anyone is interested.

#
# bucketbrigade is a rule-based system for creating a subset of a package pool.
#
# The bucket.d directory contains a set of executables, each implementing a
# rule.  Each executable receives the current working list of packages on its
# standard input, does whatever processing is necessary to fulfill its rule,
# and outputs the result to its standard output.
#
# The output of each script can be either binary packages or source packages,
# or a combination of both.  Rules may emit any kind of package (virtual, meta,
# etc).  They may also emit expressions (exim | mail-transport-agent, for
# example).  Just keep in mind that the end of the chain will emit a list
# of source packages, so you'll need rules to evaluate virtual packages and
# expressions if you generate them earlier on.
#
# As an example, lets take the following rule chain:
#
#   essential | standard | vimsucks | emacsrules | depends |  > pkglist
#
# The essential and standard executables would probably just calculate the
# Essential: yes and Priority: standard package lists, respectively, and
# write those lists to standard output.  They would also pass their standard
# input unmodified to their standard output, because they don't care about
# packages added by earlier rules.
#
# vimsucks would dump every line on its standard input to its standard output,
# except for any line that says 'vim'.
#
# emacsrules would write its standard input to its standard output, and will
# tack emacs21 to the end of it (or beginning, order doesn't matter).
#
# depends would analyze each package on its standard input,
# calculate its dependencies, and write the resulting set of packages
# to its standard output.
#