Hello,
My name is Ionuț Arțăriși and I am a student in Bucharest, Romania. I have written code for pkgdb over the last year under the supervision of Toshio Kuratomi and I am now applying for Google Summer of Code 2009.
I have already discussed this idea at length with Toshio, but I would be very interested in your ideas and comments as well. I hope to be able to implement this project as I am aware of its long history as an idea.
Here's a short description (for the more detailed/technical part please visit [1]):
Other *nix distributions have a centralized place which enables the user to search and find relevant information about a certain application (ex: packages.debian.org, freshports.org, aur.archlinux.org etc.), but this is missing in fedora. The pkgdb is developer-centric as are other related applications which hold this information (bodhi, koji, bugzilla etc.).
The proposal is to create another view of the pkgdb aimed at the general fedora user. The project will be integrated with existing tools (yum, bodhi) with a desire to centralize their information and present it in a suitable form for the intended audience. It will provide package information such as: project website, pkg version, occupied space, categories, dependencies, user commenting/reviewing of packages, user tags with keywords (which will be exported to other tools in the future and used for searching).
Thank you!
On 03/25/2009 06:03 PM, mapleoin@lavabit.com wrote:
The proposal is to create another view of the pkgdb aimed at the general fedora user. The project will be integrated with existing tools (yum, bodhi) with a desire to centralize their information and present it in a suitable form for the intended audience. It will provide package information such as: project website, pkg version, occupied space, categories, dependencies, user commenting/reviewing of packages, user tags with keywords (which will be exported to other tools in the future and used for searching).
Not to put a damper on your idea, but this is precisely one of the goals of Fedora Community. See: http://www.mail-archive.com/fedora-announce-list@redhat.com/msg01493.html
That email doesn't put it very well, but I think you'll find that it will cover a lot of the bases you're interested in. We're hopeful to have something live within the next few weeks.
Once we've got that up and running, there will undoubtedly be functionality that is in your list that we have not provided yet, and I think adding those apps to the Fedora Community environment will be a great GSOC.
~spot
Tom "spot" Callaway wrote:
On 03/25/2009 06:03 PM, mapleoin@lavabit.com wrote:
The proposal is to create another view of the pkgdb aimed at the general fedora user. The project will be integrated with existing tools (yum, bodhi) with a desire to centralize their information and present it in a suitable form for the intended audience. It will provide package information such as: project website, pkg version, occupied space, categories, dependencies, user commenting/reviewing of packages, user tags with keywords (which will be exported to other tools in the future and used for searching).
Not to put a damper on your idea, but this is precisely one of the goals of Fedora Community. See: http://www.mail-archive.com/fedora-announce-list@redhat.com/msg01493.html
That email doesn't put it very well, but I think you'll find that it will cover a lot of the bases you're interested in. We're hopeful to have something live within the next few weeks.
Once we've got that up and running, there will undoubtedly be functionality that is in your list that we have not provided yet, and I think adding those apps to the Fedora Community environment will be a great GSOC.
I disagree. I think this is a great thing to do this year in the PackageDB. When Fedora Community is up and running, giving access to the data via the Fedora Community platform could be a great project for next year's summer of code (or over the course of the year if maploin wants to continue working on that aspect). There's a few reasons for this:
1) Fedora Community pulls data from other places and puts them together usefully. This is information that belongs in the PackageDB. Exposing the information via Fedora Community is not precluded by this.
2) In fact, this work is going to be essential for helping Fedora Community to do its work. Currently, Fedora Community doesn't store information in the databases itself. It uses the data from the other Fedora Services. Having packagedb changes available via RSS feeds, tag metadata, and a search interface for the tag metadata are useful for Fedora Communityas well as PackageDB.
3) Quite a bit of this work needs to be done anyway. Tagging of packages in the PackageDB is a prerequisite for the redesign of how yum works with groups. Fedora Community is structured to primarily be a consumer and aggregator of information rather than a provider of information to other apps. Writing code that generates the tag data from the packagedb flows naturally from the other information that the PackageDB stores.
4) Fedora Community is continuing to evolve. As far as I know, the first version is trying to focus on developer UI. This is going to work on adding end user UI to an existing stable service. Even if we build a Fedora Community interface on top of this data for next year's GSoC we'll still have benefited from having users using the existing interface and suggesting enhancements that we can take into account when designing the next generation interface.
This is necessary work that will pay dividends for Fedora Community as well as the PackageDB.
-Toshio
On Wed, 2009-03-25 at 17:05 -0700, Toshio Kuratomi wrote:
This is necessary work that will pay dividends for Fedora Community as well as the PackageDB.
There is some unfinished prior art here: https://fedoraproject.org/wiki/Features/ApplicationInstaller It may not quite be the same, but the overlap seems considerable.
Matthias Clasen wrote:
On Wed, 2009-03-25 at 17:05 -0700, Toshio Kuratomi wrote:
This is necessary work that will pay dividends for Fedora Community as well as the PackageDB.
There is some unfinished prior art here: https://fedoraproject.org/wiki/Features/ApplicationInstaller It may not quite be the same, but the overlap seems considerable.
<nod> The web application portion of that feature was an influence in this design. It's too bad that was done as a separate application instead of integrated with the PackageDB otherwise we might have been able to carry the work forward when the lead developer stopped work on amber.
-Toshio