I hacked together a small script which reads the Packages.gz files from a configured set of repositories and generates a catalog page showing all packages, grouped by sections. Maybe someone finds this useful.
There’s an example page and the script source. One thing you may notice is that there’s little consistency between packages currently.
Personally I think any application catalog should be able to handle and collect the mandatory information (sections, name, version, description) automatically so that the data is taken straight from the packages where it has to be anyway. That should encourage using repositories instead of links to files as well as increase package metadata quality. Would just need a way nice way to integrate the package data with screenshots, comments, etc. Maybe Debian or Ubuntu already have one, I don’t know.
Of course once you crawl multiple repositories like this you could even mirror all the packages in one repository and save the user the trouble of configuring several of them in the application installer.
View Tommi Komulainen's profile