Me and Conna were talking about potentials for Ucurl tonight (He's really not that bad of a guy, okay?) and we started edging towards a UGet-functionality. Then, I was struck with inspiration; there's a very simple way we can half the work involved with UGet while saving us bandwidth as well.
Put simply: It's basically the same system, but instead of our server containing the repository information and the actual files, people will make their own repositories and give people a text file to stick in a folder (IE, uget/repos). UGet will read all these files and hook up with the repositories. These repositories can be large or small... only containing a few scripts (IE, a Ulysses repository that has ULib and ULX) or they can be mammoth conglomerates. In light of this, we'll need a way to deal with duplicates across repositories, but that doesn't seem like too big of an issue.
I realize this opens up more security issues, but I'm going to implement protection on Ucurl so it can't write outside the garrysmod folder... and people probably won't be connecting to untrusted repositories anyways.
Conna's volunteered to do a gui for this project if we go through with it, and we were considering something similar to how the scoreboard works. There will be important information on each line for each package, and you can expand it to see more details as well as buttons to update, disable auto-update, etc.
What are your thoughts?
EDIT: Just to add... by releasing Ucurl, I realized that there is most certainly a need for something like this in the gmod community. I had two different developers approach me separately asking for permission to use Ucurl in their scripts as an auto-updater. Now is the time!
