Pau Garcia i Quiles pgquiles at
Tue Feb 22 13:20:45 CET 2011

2011/2/21 John Layt <johnlayt at>:
> On Monday 21 February 2011 14:46:24 Pau Garcia i Quiles wrote:
>> I have added a couple of ideas for Google Summer of Code.
> "Your task: create a new installer, something which looks much more
> appealing and easier to use. The Apple Mac AppStore is a good example."
> Might I suggest that adapting the KDE/Linux App Store part of Project Bretzn
> would be a better approach? Imagine the same app store front-end for KDE
> (and potentially other FOSS apps) across Linux, Mac and Windows, using the
> OCS and all the other bling it has?
> Using the Qt version of the store gui would probably take minimal effort to
> run under Windows, and the maintenance and enhancement side of the gui would
> then mostly belong to someone else.
> My understanding is that the app store uses PackageKit to abstract the
> different package management systems the distro's use, so the primary task
> would then be to adapt the KDE Windows installer code to be a PackageKit
> backend.
> Worth a think about anyway.

Yes, I had thought about that too. In fact, I talked about that with
Frank and Leinir at FOSDEM.

However, there are a number of problems:

- Bretzn uses OpenSuse Build Service. The only way to produce software
for Windows would be cross-compiling using mingw. Leinir thought OBS
offerered Windows VMs with Visual C++ but Frank told me he thinks it
does not. I've googled and it seems Frank is right :-(
- If I have understood Bretzn right, it comprises infrastructure,
procotols, workflow, etc but not an application. We still need an
application (our current installer, or a new one).
- Dependencies: Bretzn does not really need to go too much in depth
with dependencies because on Linux, BSD, etc that's not an issue. On
Windows, it is because there is no package manager.

So yes, I definitely like Bretzn, but I think it will not be possible
at this moment (I hope it will be in the future). Further: the ideas I
proposed for GSoC are not incompatible with Bretzn. I think we need to
go for all of them:

- Improve emerge (I have something more to say about this but it's for
another e-mail)
- Improve the installer or develop a new one, with a more appealing UI
and better HMI. I think most of the lower-level code (resolving
dependencies, downloading stuff, etc) can be reused from the current
- Then, integrate the appealing UI with Bretzn, once we have

In fact, I think the "appealing UI installer" could be also used by
KDE Mac (at least the UI). They currently rely on MacPorts, HomeBrew,
Fink, etc which are not exactly user-friendly either.

Pau Garcia i Quiles
(Due to my workload, I may need 10 days to answer)

More information about the Kde-windows mailing list