its not a webservice :P
Only generates a xmlfile with info of packages
<Package>
<name>paprefs</name>
<version>0.9.7-0.1-1</version>
<shortDesc lang="en">Sound Server Preferences</shortDesc>
<longDesc lang="en">View and modify the configuration of the local
sound server</longDesc>
<categorie lang="en">Settings</categorie>
</Package>
this xml file it called cache. Conary not support search. But with the
cache can doing search by name, *Desc, and Search by Categorie.
So packagekit its very slow because when do a refresh it fetch all
packages( reqquery ) and insert on sqlite3 database. more than 5341
entries on a slow database.
So for implement the New Cache its 35% of process i have a new XMLCache
what can doing search.
BTW i want add support for multiple repository. read this:
http://lists.rpath.org/pipermail/foresight-devel/2009-January/000710.html
( the resolve command its done )
But present some problems with conary because im lost on the update
remove process :S so for this make a conary-lib (
http://lists.rpath.org/pipermail/foresight-devel/2009-February/000711.html )
With it replace some part of code of backend and show something more
easy to understand. For next PK hackers. And use on other projects.
Ken VanDine escribió:
> This is exactly what zodman has done :)
>
> He has created a web service that creates a xml file with all the
> package metadata, and next up is making PK fetch that xml file for
> refresh cache. So refreshes will be super fast and lower the repo
> load. Plus this gives us a way to include more metadata to make
> package browsing possible.
>
> --Ken
>
> On Thu, Feb 5, 2009 at 8:23 AM, Stephanie Watson
> <s...@foresightlinux.org> wrote:
>> I just had one of my
>> checked-email-first-thing-out-of-bed-and-had-what-seemed-like-a-great-idea-at-the-time
>> moments. Note that I am not currently up-to-date on everything folks have
>> been doing with PackageKit on FL up to this point, only that there are still
>> a lot of usability issues to work out.
>>
>> If everyone who uses PackageKit is querying the same repository, and Conary
>> repquery is the reason browsing available software is slow, my idea to work
>> around that issue is to have something in the middle (from FL, not running
>> on the client) that PackageKit is fetching instead of running a full
>> repquery, and have that middle thing doing all the requery work in the
>> background constantly running a new requery each time there's a commit to
>> the repository. This middle item would simply me a static index of
>> repository contents that is being constantly updated, but that is an
>> easy-to-retrieve format like a CSV file.
>>
>> PackageKit <---> Index of Repository Query Results <---> Conary
>> Repository
>>
>> If the user selects items in PackageKit after that, the new conary update
>> commands can still be formed and executed against the repository at that
>> time.
>>
>> Thoughts? Has anyone considered this approach before? Any confirmed
>> sticking points?
>>
>> I'd like to try my hand at planning and coding something like this, but I
>> would need time to get familiar with the PackageKit code, too, and have an
>> understanding of the approaches people have been taking so far.
>>
>> -Stef
>>
>> --
>> Stephanie Watson
>> Raleigh, North Carolina
>> _______________________________________________
>> Foresight-devel mailing list
>> Foresight-devel@lists.rpath.org
>> http://lists.rpath.org/mailman/listinfo/foresight-devel
>>
> _______________________________________________
> Foresight-devel mailing list
> Foresight-devel@lists.rpath.org
> http://lists.rpath.org/mailman/listinfo/foresight-devel
>
_______________________________________________
Foresight-devel mailing list
Foresight-devel@lists.rpath.org
http://lists.rpath.org/mailman/listinfo/foresight-devel
