Le 01/10/2010 12:38, Fabrice Facorat a écrit :
I've been following closely all the Mandriva vs Mageia story. I found
it unfortunate that we have to come to this way, but I guess there's a
serious fracture between Mandriva and part of its community. We have
no choice except to cope with this and try to do our best to allow
this unfortunate situation to found a sensible solution in the future.

As we know, one of the Mandriva strenght are the Mandriva tools,
however Mandriva tools have some issues :
- they are written in perl. Sorry for perl dev, but I do still think
that perl is harder to understand than C-like based syntax langages.
However we must admit that we are not going to rewrite all the
Mandriva tools ;-) However better documentation ( PerlDoc tags ) could
help a little.

- Mandriva tools are not used by others distributions ( except
PCLinuxOS, United Linux, and ... Mageia ) and so have few external
contributions : They notably lack visibility.

I do agree with this.

I do think also that Mandriva will have to use its ressources in an
efficient way.

Here aree my proposals, feel free to discuss :

1. host Mandriva tools on github or code.google.com. This will ease
fork maintenance and tracking, to contribute back ( without having to
have a Mandriva account )

Yes. Having their own 'site' and independent platform may help.

2. Make some decisions about the tools we should keep, and the ones we
should ... trash. For example we did replace printerdrake with
system-config-printer ( python ), and msec have been rewritten (
python ). Whereas I do think that system-config-printer is way buggier
than printerdrake, I guess that at some points, we will have to do
this more and more : replace some Mandriva tools with for example some
Fedora ones. Please note however that this bring its own issues :
python vs perl, and the integration with the rest of Mandriva
infrastructure

We need to see what is still functional, what is broken (and so what is to repair), and what is to drop. Eventually, what is to support and repair again (printerdrake?) if possible.

For what I know, there are many tools that work : RPMDrake and related, Drakstats, Diskdrake, Harddrake, DrakX11, Drak3D, DrakUPS, DrakFirewall, DrakGuard (wonderful this one) but may networking tools to share network or use VPN, Samba, NFS, WebDav, and eventually along with Diskdrake, are broken. Others such as Draksnapshot and DrakSamba (not sure if it works or not) are a pain due to insufficient functionalities or outdated/painful GUI. There is a nice theming functionality in the MCC that is also probably broken or difficult to use, that could be restored or explained.

3. A decision will have to be made concerning net_applet and NetworkManager

Yes, even though I think we should give another chance to NetApplet and see what should be fixed to make it better. NetAppler has the advantages of being linked to DrakFirewall, perhaps other tools, and to be independent of any environment.

4. Whereas I do love rpmdrake, I do think also that something will
have to be done about it as its UI is clearly outdated and not on par
with the competition :
- Ubuntu software center :
http://seilo.geekyogre.com/2010/09/software-center-with-a-dose-of-zeitgeist-and-maybe-teamgeist/
, http://en.wikipedia.org/wiki/Ubuntu_Software_Center ,
https://wiki.ubuntu.com/SoftwareCenter
- iTunes App Store :
http://www.askdavetaylor.com/how_to_download_iphone_apps_from_apple_itunes_store.html
, 
http://cybernetnews.com/download-iphone-firmware-20-itunes-77-app-store-and-more/
- Interesting discussion about PackageKit direction :
http://mairin.wordpress.com/2010/09/01/a-story-about-updates-and-people/

So we may have to completely rewrite rpmdrake UI or switch to
packagekit with and urpmi backend.

I still have a very strong faith and appreciation for RPMDrake. I really think it is well designed and intuitive, despite it's little issues and being slow (honestly, PackageKit is slow and also has issues so...).

The real issue that RPMDrake has is it's Aplications with GUI filter. Even if I think this functionalitiy is really good for beginners, RPMDrake is a -package- manager. Mandriva does need a real and dedicated Application manager (could be called an AppCenter) where beginners would find a way to install (shop?) applications with a very nice layout, presentation, clear icons, screenshot, and no irritating choice of hundreds of dependencies with barbaric names. It might be difficult, but much more convenient for those who just want things to work in a snap (or in very few clicks).

This would allow to place back again the default filter on "All" (should be renamed to "Show all packages") for the RPMDrake package manager. We would then have an AppCenter and a real package manager for advanced package management, without forgetting a dedicated tool with GUI to manage orphans more efficiently.

Yet, all of this demands a huge lot of work and we will need huge resources...

5. Junior tasks contributions. I noticed while visiting the
LibreOffice website. They have junior task for people willing to
contribute to the codebase, and most of theses junior tasks consist to
improve code clarity, fix comments. I guess that the same thing could
be done with Mandriva tools, notably adding perldoc tags/comments.

Yes. This will help people from outside understand better how the program works. Reading the code itself isn't that easy even if it is well written.

Last but not least, I know that on Mageia ML, there was a discussion
about the people we should target. Here are some interesting
reflexions :
Sweet Caroline : http://mairin.wordpress.com/2010/09/02/sweet-caroline/
fedoraproject.org redesign update :
http://mairin.wordpress.com/2010/09/03/fedoraproject-org-redesign-update/
You must be this tall to ride: __ :
http://mairin.wordpress.com/2010/10/01/you-must-be-this-tall-to-ride-__

Will read them when I'll find the time to...

Thomas.

Reply via email to