Regarding GSoC Plasmoid data caching project

[todd rme]

On Tue, Mar 30, 2010 at 3:13 AM, shruti jain <shruti.jain1988 at gmail.com> wrote:
> Hello everyone !
>
> I am a final year student of computer science. I intend to contribute to
> plasma and see GSoC 2010 as a good starting point for that. I have
> participated in GSoC 2009 as well. I am proficient in C/C++ and network
> programming as well. I have also been the system administrator of my
> university.
> I am interested in Plasmoid data caching project. I am doing research on
> speculative execution at my university. It has a similar concept of starting
> execution on cached data while the most recent copy of data is not available
> from the server and avoid downloading from server if the data has not
> changed since it was cached.
>
> Presently, plasmoids such as news etc do not show anything when Plasma
> netbooks are disconnected from the network. The idea is to use the cached
> data when network is not available and refresh when network is available.
> I think that the project can be extended to incorporate collaborative
> caching as well. Even if the network is present, we can take the data from
> caches of other clients on the network. This reduces the load on the server
> from where the data(like news) is being fetched. This also fetches a more
> recent data in case of proxy failures.
>
> Please send me your comments on the idea and any suggestions to improve the
> same.

I am not an expert on plasma, but my suspicion would be that, when
possible, this should probably be done at the dataengine level, rather
than the plasmoid level.  That might be able to prevent duplicate
caches when multiple widgets look for the same data, and will allow
many plasmoids to get the caching with little or no additional effort.
 This would also allow things like the weather wallpaper to show a
wallpaper when there is no weather, which I think uses the same
dataengine as the various weather plasmoids.

Speaking of which, if it isn't already implemented already (it may
be), it might be nice if various web service data engines that are
looking for the same data (like different weather engines looking for
data from the same location in the same provider) would all share the
same data rather than downloading it independently.  As I said, this
may already be how it works, I haven't looked into it.

-Todd