The main problem I see with adding them all is if one repository is down the update will fail and you'll need to track it down through the log.

I don't have nearly that many setup on mine. I'd suggest just adding them when you find an application you'd like to install. You can check the old applications catalog or the new applications database for what's available.

To solve this, we use a python script that searches the repositories of sources.list and checks if they are down.

http://www.todosymbian.com/postt31284.html

Also, I would like to make a script that searches all packages listed in sources list, download to my home PC, make a new packages.gz and use them offline.
Can anyone help me to make this?

I only spent a few minutes searching around for something that might work. It appears there are a few packages that may do the job and build a local mirror of the repositories: apt-catcher, apt-proxy and apt-mirror. Here's a howto on creating a proxy and this one for creating a mirror.

I don't have a linux machine handy right now, but I'd bet the apt-mirror would work.

I like the idea and if it works we could actually create a single master repository for packages that could be hosted at several different locations for speed and availability.

I booted up the maemo ubuntu appliance to check it out and apt-mirror seems to work ok. You'll need to update the /etc/apt/mirror.list sections to change the architecture, mirror location and to add in the repositories you want to mirror.

# mirror.list config changes
set base_path /var/spool/apt-maemo
set defaultarch armel

You'll need to create apt-maemo if it doesn't exist. Copy apt-mirror for a quick start.
copy -R /var/spool/apt-mirror /var/spool/apt-maemo

# The repositories are the same format as listed earlier in this thread and in your sources.list.
deb http://scriptkiller.de/apt/ mistral main
# etc, etc.

To access them, you'll want to setup a web or ftp server to access the mirror. Then, update your sources.list on the nokia to reference the repositories.

for example, this should work (I haven't verified yet, though) assuming you're pointing apache to the /var/spool/apt-maemo directory.

http://local.workstation/mirror/scriptkiller.de/apt/ mistral main

(Using the sources.list earlier and adjusting for broken sites, there was about 650MB or so in repositories. I mirrored scriptkiller and maemo-hackers fine.)

Check out the howto linked to earlier for more details. I had it setup in just a few minutes. I hope this helps!

Thanks, it seems interesting.

I tried all that you said, and it seems to work.
Well I think I have to make some improvements like to download all the stuff in a single directory and make a new packages.gz to put all of them in a single USB stick
But, as I'm a newbie in linux it is quite difficult to me.
Also it would be interesting to make a script to make all of this in a simple click.
Can anyone help me?

Thanks

I started on this but ran out of time after I hit a snag. However, you should be able to download all the packages using apt-mirror and then copy the .deb files to a single location and create your own repository from those. There's a repository howto here.

As long as you copy the .deb files you can use apt-mirror to refresh the packages without having to download them all again.

One thing I found is that you may have package version conflicts between mirrored repositories. The other was that I could connect to the new web repository I created, download the package but it didn't install for some reason. I'm sure it's because I didn't have the repository setup quite right.

On a related note if any developers are reading, would it be realistic to have a central repository where developers or maintainers could upload packages to an incoming location and have a process pick it up, validate and add it automatically? I could even see this being tied to the application database similar to the way debian does with their packages. That way you select your application from a single repository and shouldn't need to add in 15-20 repositories to get what you want.

Interestingly enough, I don't see any information in the maemo database that isn't already stored in a package. Building the database from the repository seems like the logical way to do it and not building a database from multiple repositories and relying on someone to manully keep them in sync.