0

When updating an installation, it makes a lot of sense using apt-get-cacher if you have more than one computer to update. The more PC's that needs to be updated, the better it is to have a local cache.

Today, you need to enter a rather long and complicated command to set apt-get to use this proxy/cache and the same if you want to stop using it. Especially for someone that travels with a laptop, it would be useful to turn this on and off in a convenient way.

It would also help anyone that is thinking about using this software to implement the use.

I suggest that an option is added to Kpackagekit where you can specify the name/IP of the caching server.

This is a very distribution-specific idea. Only debian-based systems use apt. I could understand the packagekit backends providing configuration option to packagekit, but I think that is the responsibility of the backend developers (the distribution) rather than packagekit (once packagekit provides an API). And this is definitely not a KDE thing, since KDE simply uses packagekit. If packagekit does not support it then kpackagekit cannot, either.

What about letting to save/export downloaded deb or rpm files instead of using some virtual cache.
This way you could just put these files in dir and share over network or even create a mirror.

This is the problem of normal users. We see a problem, and we suggest a solution to the contacts we know upstream. And I can absolutely understand that it might not be a specific KDE idea when it is being explained. It is not easy for a normal user to understand and have any idea about these things. I think this might be something that could be looked at as a fundamental thing with the brainstorming section. Make a system that make it easy for you to pass something on to the right people...

For a normal user, everything that look like a part of the OS, is a part of the OS. To explain that KDE is not a part of the OS is wasted energy. So a lot more of this type of suggestions will come in the future. A normal user might feel like this is like hitting their head on a brick wall. My suggestion:

Mark a suggestion like this "3rd party responsibility", but let it at least be discussed as an idea. As it is now, I have no idea where to go to suggest the idea, I will search for packagekit. But my experience with some developers is that they are not always easy to get hold of. So even if the idea is good, it might not get there just because of unread mail or missing mailaddresses. I guess since KDE uses packagekit, you have a good connection with the people behind it. So why not pass things on? It carry a lot more weight if it comes from you

Lukas wrote:What about letting to save/export downloaded deb or rpm files instead of using some virtual cache.
This way you could just put these files in dir and share over network or even create a mirror.

I am running this cache at home. We have 4 pc's running Kubuntu and it saves a lot of download time. Your suggestion COULD have worked for me IF:

There was a interface to export.
That interface was able to detect already exported packages so nothing would be done double.
There was an interface to import.
And probably some more...

I a school or an organisation with many more systems, there are some weaknesses that start to come to light.

How does an organisation with 1000 computers keep their system updated in an easy way? There absolutely should be a way to use a proxy/cache for this, but you might want to restrict it more - ONLY install from cache. This way, only approved packages are installed. An organisation might choose to hold back updates for different reasons.

So not only should there be an easy way to set this, there should be a good interface for the cache where you get a constant list of new additions and a checkmark to make it available on the LAN.

I like what Mint has done with their updates, mark them with a number so you know what is REALLY safe.

Yes, I know this is way out of the KDE area. But it might still be the best place to discuss it and get feedback on it from an end user perspective...

I haven't investigated how exactly repos work, but if I understood right, lets say http repo is a folder with debs/rpms to install and a file with info about these files.
The best scenario could be an option in package manager to export files in a specific folder and create/generate repo description file. Then install apache or any other http server, add it to repos list with highest priority.

When any network PC would try to update, they will first look at local repo, and if this repo has newest version - great use it. If there is a new and not yet cached update, or an specific app for that only pc, network PC would download installs directly.


p.s. does using proxy in your way forces to use ONLY data from proxy, or it also looks up in repos for updates?

Another idea, all we need is just a apache (or any other http server), 1 .htaccess file, a ~200 lines of php/phyton/RoR script, a corn do delete old files.

Very similar as any www proxy servers with cacheing feature so it could precache all packages on the fly. It wont be a difficult script to write.

No need to touch pacakgekit in any way

The only possible problem could be signing data...