Hi,

> Yeah, standalone is better. It's easier to maintain for all of us who
> aren't really perl experts, and honestly avoids the mess that was
> drakxtools.
Well honestly not all the drakish has been removed, not easy and since
we are just two developers atm, we would not have had any demos to show
that this project could work :)

We chose perl just because we thought we could have more helping since
all was written in perl, unfortunately we were wrong, but i learnt perl
focusing on the modern aspect of the language and with Moose the OO
as well.

> I know that libyui doesn't use xembed (I know I was supposed to PR an
> implementation a long time ago -- oops) - have you found a replacement
> for that for AdminPanel, or are we just launching the standalone
> module?
I din't know xembed :) interesting thing, but i suppose it does not
work on console (single user mode) and so using just ncurses, am i wrong?
mpan is a launcher, but if perl modules are written following the rule into
http://gitweb.mageia.org/software/adminpanel/tree/MODULE_HACKING
(hoping is uptodate :) ) it just instantiates and run them by using their
start method.
So it's not just a simple launcher, but it needs to be when not using perl,
or embedded modules (rpmdragora for instance).

An additional note must be said on mpan, as said before it's more a demo
than a real final project, and i think it needs more work to be
improved. xml configuration file does not manage translations for instance
and maybe a better layout could be studied instead of having an icon button
collection, as category layout....

libyui also has some limitations, even if most can be avoided by using
some other widgets, so having also libyui hacking helps, to provide
fixings and improvements to be used on my branch and hopefully upstream.

>> Embedded modules are in Perl, but external could be written in
>> c++ native, or python, mono and ruby also, using the libyui-binding.
>> We added some new widgets so also libyui-mga plugins. Some improvements
>> are not in upstream master and used my branches on github.
>
> I saw that - since they're all standalone, and the admin panel simply
> launches the corresponding module, it works for me if OpenMandriva
> needs to add any custom modules.
Yes, the main idea however is to write a common (Shared) module that
aim to work as or with back-ends, so that the GUI part (Module) can be
used just for presentation layer, now implemented by using libyui.

That should also allow to use the Shared part in other situation such
as our installer if we don't want to use libyui there for instance.
So services shared api for instance are into

AdminPanel::Shared::Services
(note that AdminPanel ns should be changed asap)

while the gui implementation

AdminPanel::Module::Services

and we strongly avoid any dialogs using in the Shared part
(just into Shared::GUI perl module and on for the common used ones)
that approach could allow to split packages to focus to the module aims
in future if needed (i.e. Shared, Shared::GUI, Modules,...)

Moreover rpmdragora could use PackageKit if someone really fixes urpmi
backend there, and with packagekit not only urpmi is managed, increasing
the cross-distro aspect of the project, that fwiw i made run also in
debian wheezy, just manaclock and manahost really, and something else
with not so good results :)

>> Let me know if you like to participate :)
>>
>
> Yeah, sure, I'd be interested!
Great! Feel free to contact me also privately to have feed backs on what
you need to start and which kind of help you can provide :)

Cheers,
   Angelo


_______________________________________________
OM-Cooker mailing list
[email protected]
http://ml.openmandriva.org/listinfo.cgi/om-cooker-openmandriva.org

Reply via email to