Re: Massively Overloaded Linux Database?

*This message was transferred with a trial version of CommuniGate(tm) Pro*


> Ok, the consensus seems to be that a large database of
> software would be more useful than the software itself
> on CDs.
> 
> I can buy that argument. Freshmeat is OK at this, but
> you can't do contextual searches (yet) and you can't
> tell apt-get or rpm to fetch binaries through it.
> 
> What's needed, if I am understanding this correctly,
> is a cross between "archie" (anyone remember that?),
> freshmeat, google, squid and an apt server, with a
> local cache of rare and frequently-requested software.
> 
> The archie part would enable the server to identify
> who had what binaries for a given program.
> 
> A Freshmeat-type database would give you a list of the
> most recent versions of a program, given the title of
> the software, rather than the filename.
> 
> A Google-like search engine would give the user the
> ability to context-search. There is a lot out there,
> and you can't assume the user knows every single
> program by name, but they know the context in which
> they'll be using it.
> 
> A squid-type proxy allows you to handle cached
> queries, forwarding those that can't be processed
> locally.
> 
> The apt-server would be the wrapper around most of
> this gunk, so that retrieved data can be returned in a
> format that apt-get can handle.
> 
> All the components are easy enough to obtain, but
> yeesh! I don't think anyone's really looked at
> combining them in quite this way.

This is exactly what I was suggesting.  Good job at being a little more
precise than me! :)

I don't believe that it has to be done through apt-get though.
Why can't it just be http download?