I'd like to make a website called AptOffline (ServerAPToffline) or something similar with the purpose of being able to download programs for ubuntu or debian (and also rpm) in and EASY way, via WWW; this is to avoid the difficult research, if not impossible sometimes, that a user without internet line has to do from an Internet point in order to install a program, considering all the necessary or not necessary dependencies, and the correct versions for the system in use.

The IDEA is to have a web interface where the user can upload the file containing the list of the installed programs (DATABASE) on the offline machine.
From this point one should be able to use the site as if being using synaptic (or aptitude or similars), allowing thus to:
-choose any programs with the calculation of the dependent programs o libs (and only those not already installed as identified from the DATABASE)
-to have an output, instead of the installation procedure, the list of the programs in form of links to download or, better still, one unique zipped file containing all the files already grouped from the repositories
-eventual commands to uninstall the packets no more necessary before or after the update.

I an not a programmer and I know not enough about Apt funtioning, anyway what I have with many other user is the necessity... so:

-is it necessary to have a linux web server with apt istances to run underneath or, can this be all done via database queries?
-Which is/are the DATABASE files the user has to collect for the upload?
-the ServerAPToffline should be constantly updated with the latest packets lists from the main repositories (the more, the better)
-how to simulate the Synaptic Software via web showing the list of available and installed packages? Being able to select them , apply and calculate dependencies
-a user might have an account in this site where one's own DATABASE file is kept and is update from the user ONLY when changes to the installations are not identical to those made on the site.
.............

Ciao

reasonable idea i principle. at LQ we certainly have a lot of issues with people without reliable internet connections. the biggest problem you'll have as i see it is that without knowing *exactly* what packages are already installed on the client system in question, it's futile. if you suggested *every* dependency, then you'll go back to glibc and a kernel every time. you're never going to be able to rely on a generic reference site, possibly unless you go from base installs and such, but that could lead to vast amounts of unnecessary packages being installed. you would also really need more information that the name of the package. many rpm's require "provides" attributes, filenames typically, e.g. a specfic version of a glibc binary, not just the name of a package.

Chris,
That's why I have though that one should bring from his home linux the DATABASE file with him, or better have an account that keeps record of all of the changes on the site...

Where else can I send this in order to have better audience?