Debian System Wide Information Manager
You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
 
 
 
 
 

126 lines
5.7 KiB

* Configuration for --dbpath, --root, --main, --contrib, --non-free,
--non-us, --arch, --dists could be automated so that a state could be
established on the command line, without needing to enter any of these
options again, untill a new state was desired, --default would bring
everything back to the default directory /var/lib/dpkg.
* Non-root users could run all functions if permissions were changed,
probably this should be left up to the system administrator.
* Add --todo --bugs to -q.
* Looks like --source and --source_only is kind of deprecated now that apt_0.3.6
is out, and there are Sources.gz in the appropriate places. But, now
there exists three unique ways to get sources, debget which doesn't use
databases, apt which uses databases, and swim which combines the
approaches of both apt and debget, all three ways have their advantages.
Swim isn't dependent on the existence of an up to date Sources.gz.
* --ftp automatically removes packages with changed upstream-revisions..
perhaps move them to a storage location.
* base-files package has changed it's file/dir listing, eventually
SWIM::File can be modified in the future.
* SWIM::Apt needs to be modified so that when it reads the Packages names
in the default directory it will look at the basename to take into
account non-Debian distributions. Non-Debian distributions should
be presented separately when DF is used as an argument. Probably a new
option will be added. This adheres to the swim philosophy of allowing
independent distributions with the exact directory structure of a normal
Debian distribution to exist.
* mention netselect to help people find the best mirror site, and maybe
automatically use netselect on swimz.list to find which sites are up,
and are best to use first, overridding the default sequential order!
* Yes, you can 'cp -a' and 'gzip -d' sources from swim to sources for
apt, then just do a swim --apt --check or vica versa.
* Make sure this doesn't happen for proposed-updates - "swim:
dists/proposed-updates/single/binary-i386/Release does not exist on the
server", and make a note of proposed-updates in the manual, no Contents.
Incoming can be grabbed by doing something like this, "deb url
debian/Incoming" vs experimental "deb url/debian project/experimental/".
Packages are available here, too, no Release or Contents. swim is too
hardwired and this needs to be changed.
* For now, a trade-off will be made in respect to updating the n* virtual
filesystem, even if Contents hasn't been updated, the versions of C,CR
will still be updated with the old Contents, on the other hand if
nfile* was version-less this wouldn't be necessary, then the behavior
mentioned next could be implemented anyways. If Contents are ever
updated on a more consistent basis, this behavior will be changed to
wait for these packages untill the next Contents, but basically this
should be done anyways, it will mean, quick updates during the week, and
one long one when the new Contents becomes available, the lack of
versioning would make this viable. Basically, a database will be kept
for .packagesdiff-arch-dist.deb, nfile* will remain versioned because
this means faster lookups.
* Implement the "on the fly" distribution creation system.
* Add the DF virtual system which will allow --apt2df and --df2apt to
work.
* a search for everything found in -T
* -o DPkg::Options::=--purge once --remove isn't a standard option to dpkg
for "remove", and possible --admindir (like --dbpath) --root (like
--root) and --instdir (where packages should be installed).
* add --force-* at least for --purge and depends.
* an option which would show all packages relationships for a package
being --ftp or -xyrz on the --stdin line
* Complete menuindex() which allows a search of the debian menu system.
* add the pid (personal information database) editing, and data storage.
* Use the MD5 module from libmd5-perl for --md5sum?
* gnupg, pgp, and md5sum checking for source probably using dscverify from
package devscripts, this uses the MD5 module
* Need to add the capability to deal with weird non-standard Filename: .*
situations, both in retrieval, and archiving. In situations like this,
the distribution may not be apparent, but the package could still be
found and placed in a special area, and a special database could store
the information. This database could be transparent, and available for
quering regardless of the distribution default, or command line
specification.
* Add the --file and --http capabilities or at least the capability to
look at Packages from strange directories grabbed by apt using these
methods. Meanwhile, swim's ftp capabilities make slow modems happy, and
allow for precise downloading versus downloading every package related
to -T.
* Add a hash option so that swim can know that po -> unstable, this would
work fine, because swim shows the Release version, but the author is
against implementing this ability.
* Instead of PWD!full_path, allow multiple filenames to be quoted without
pathnames for --extract.
* .., ../, ../../, ../../file-or-dir (not ../../file-or-dir/) ability for
-qf|p works, but more complex ../ is not implemented, yet. ../ will
also follow a symbolic link, this is a feature.
* --search by dirs when in place using -f.
* Add the retro full spectrum of dpkg options, it's nice sometimes to just
do a simple install.
* Gather together all the build children into a giant family including the
teacher and the checker.
* Do the Perl for Non Free OSes thing.
* Remove a large percentage of the bold and underlining from the
documentation.
* Make needed changes and improvements to the Manual - a constant TODO.
* swim faster and faster!