yast4debian, going ahead

[Ramon Acedo <ramon@linux-labs.net>]

Hi there!

It seems there is movement in the yast4debian project, I'm glad of
it :-)

Jaldhar packaged the base packages to start working on the modules.
Great job! I will try to send an email commenting some dependencies
which should be added to some of the packages.

People who want to compile some YaST2 modules for Debian can still
follow the howto from almost the end, Jaldhar's packages are probably
the difficult-to-compile ones so come on, we encourage you to try it out
and give feedback!

Pikota is working hard in the dark and almost every day has news to
me :-) He is preparing patches to start using some modules.

He's asked me to write to the list to ask your opinion about the
matter I comment below.

There are some possible approaches to work on the yast2-packagemanager:

1 - Do-it-all approach: rewrite entirely yast2-packagemanager using 
    libapt where there is rpm stuff.

2 - Just-what-needed approach: Don't touch yast2-packagemanager, just
    hacking on the library Requires.ycp and what it depends on. As you
    know most packages need to check whether a specific package is
    installed or not. This kind of operations are done through calls
    like Pkg::PkgInstall. This approach would go through this file
    and replace things like:
    SCR::Execute (.target.bash, "rpm -q " + package);
    by
    SCR::Execute (.target.bash, "dpkg -l" + package);
    or
    SCR::Execute (.target.bash, "rpm -Uvh " + package);
    by
    SCR::Execute (.target.bash, "apt-get install -y " + package);

    But probably some equivalencies would be tricky.
    I hope to be clear enough :-)

3 - Ray approach (in terms of speed): Do not do anything, well actualy
    almost anything. We would proceed as in the 2nd approach but instead of
    looking for the apt alternative just show messages to the user telling
    things like "Portmap is not installed, please install it by hand and 
    try again later".

Pikota and me agree and we'd choose right now the 3rd approach which find
more sensible in terms of letting it just work and focus on trying to port
some other modules.

What do you think guys?

Ramon