Linux and application installing

Martin Bacovsky mbacovsk at redhat.com
Tue Sep 21 19:50:47 UTC 2010


On Friday, September 17, 2010 11:04 am, Richard Hughes wrote 
> On 17 September 2010 08:01, FlorianFesti <ffesti at redhat.com> wrote:
> > Can someone please elaborate a bit what pieces of information are really
> > needed? The .desktop files as a whole?
> 
> Information we use in app-install:
> 
> TABLE translations:
> STRING application_id Name of the desktop file, with no extension)
> STRING application_name Name in desktop file, in locale)
> STRING application_summary Comment in desktop file, in locale)
> STRING locale System locale
> 
> TABLE applications:
> STRING application_id Name of the desktop file, with no extension
> STRING package_name The package name, e.g. 'nautilus'
> STRING categories Categories from desktop file, _not_ PK groups or PK
> categories STRING repo_id For adding and removal
> STRING icon_name Local filename with extension, e.g. "totem-pl.png".
> This is required so that local theme can override upstream icon
> STRING application_name Name in desktop file
> STRING application_summary Comment in desktop file
> INTEGER rating Rating in percent. 0 is unrated, and 100% is the best
> application in the world.
> STRING screenshot_url Screenshot URL that shows exmple usage.
> BOOLEAN installed If the application is installed NB: you are required
> to run app-install-admin --refresh-installed after each package
> install or remove if you use this feature.
> 
> So, most of the data is extracted from the desktop file, but the icons
> often need to be resized and converted into png format which takes a
> bit of time. The rating comes from [is present in] comps, but will be
> populated by the gnome3 application stats, and I'm thinking of
> pointing screenshot_url at pkgdb for now.
> 
> So, you still need more data than just the desktop file. When i
> generate fedora-app-install I use the desktop data for 90% of the
> data, and then use comps to fudge a popularity score, and then
> manually add the pkgdb link.
> 
> It's a more manual process than I would like, although you can pretty
> much leave the generator to "do it's thing" which is 90% of the work.
> I think it's a ton more data than you want as rpm provides.
> 
> Richard.

I work on *yet another* application database[1] as part of Fedora package database [2].

We use very similar approach to get the data to store them and we fight the same issues.

For example now I'm working on better application detection as some packages (kdebase for instance)
contains just .desktop files for the most common applications. Also naming can be slightly different 
from release to release. This leads to confusing results like [3] or [4].

I like what you have achieved in your project and am interested in any form of cooperation.

I wonder how many other such project do exist...

Regards,
Martin

[1] https://admin.fedoraproject.org/pkgdb/apps/name/list/a*
[2] https://admin.fedoraproject.org/pkgdb/
[3] https://admin.fedoraproject.org/pkgdb/apps/search/inkscape
[4] https://admin.fedoraproject.org/pkgdb/applications/Galeon


More information about the devel mailing list