Working together on NM09 support and cleaning Solid::Control

Will Stephenson wstephenson at kde.org
Thu May 19 21:01:10 CEST 2011


On Thursday 19 May 2011 01:48:50 you wrote:
> Em Wednesday 18 May 2011, Will Stephenson escreveu:
> > To begin, 4.7.0 isn't an important deadline for this, since we can easily
> > carry out my #1 (it doesn't matter what the branch it is in is called)
> > and distros can ship that alongside 4.7.0 to support NM 0.8 with the
> > status quo. Use of NM09 in distros is more interesting and the only
> > distro to have shipped in (Fedora) has already done so and has it own
> > solution. The next round of distro releases comes in the autumn by which
> > time our NM09 support will be complete.
> 
> 	Seasons of the year is not usually used to indicate time here in the
> south hemisphere :-), autumn for you would be November, right?

Yes, and to clarify, I didn't mean that it will take until November.

> 	Just an example of seasons as indicative of time problem, in the
> northeast of Brazil what people there call "inverno" (winter in English) is
> the summer season in Brazil. "Inverno" for them is the rain season (summer
> in Brazil) and "verão" (summer in English) is the dry season (winter in
> Brazil).
> 
> > Assuming that our NM 0.9 support in the libnm-qt/solid::control/*nm09
> > layer is similar, just class renames and the new dbus API, the amount of
> > work in either approach is about the same, and leaves us with less code
> > in 4.7 to maintain than in 4.6, rather than more (in your approach). 
> > That's my analysis. Where do you see the extra workload?
> 
> 	Ok, so who would do the actual conversion from Solid::Control to libnm-
> qt? You are the most qualified for that. Could you do it? How much time it
> would take?

I'm doing it in libnm-qt branch (which is a branch of nm09 itself).  I'll add 
the libnm-qt sources there too (they are currently in my personal scratch repo 
at git.kde.org)
 
> > > I have spent more time than I should have
> > > in Plasma NM, some things in my life had been postponed to make that
> > > happen.
> > 
> > I reviewed all the commits of the last year recently and you've been
> > doing amazing work yourself and in integrating others' work.  I'm
> > personally very grateful for your personal sacrifices to make this
> > happen.  I'm aware that you're starting a new job and will not have
> > time, which is why I'm stepping up to help again.
> 
> 	I already started it. It is more like a freelance job, I can work at
> home, which by the way is in another city from where the company I am doing
> this job for is located. I have a schedule to follow, today I have
> delivered the first milestone, now I have to do the second milestone. I
> also had to stop my studyings (I am self-learning person) to spent time in
> Plasma NM. In July I am going to apply to public service (an exam, similar
> to the ones some universites do to accept students), but I have not been
> able to study during the last month.
> 
> > I don't want to go back to working on KNM alone but I do want to get the
> > codebase clean and maintainable as we add NM09 support by implementing
> > the plan from the Solid meeting.
> 
> 	We are not alone. Ilia Kats has been helping me a lot in the last month,
> he fixed the ad-hoc/connection sharing problem, implemented the IPv6
> support and fixed several other bugs. Rajeesh K Nambiar and Rafael
> Fernández López also helps me from time to time and now Lukas Tinkl is
> helping us. We just need to know how to use libnm-qt.

It's semantically the same as solid::control, just the classes are named to 
match NetworkManager's dbus API.

> > >         #1 should be based in nm09 instead of master in my oppinion. In
> > >         the
> > > 
> > > future we will need to branch master to nm08 and move the NM-0.9 code
> > > to master, but not now.
> > 
> > Ok, just a branch name, we can do it either way around.
> 
> 	Is Plasma NM able to connect to wifi using libnm-qt? If we are going to
> use it we need it commited to nm09 branch as soon as possible.

The port is not that far along yet but work is proceeding quickly.

My plan is to 
1. get networkmanagement building vs libnm-qt
2. merge divergences in nm09 branch to libnm-qt and get libnm-qt as 
functionals as nm09
3. branch master to pre09 branch 
4. merge libnm-qt to master

Will


More information about the kde-networkmanager mailing list