• March 94 - Protocols - Nice explanation of Protocols as a method for defining interfaces. Interesting to see an Apple guy (at the time) admiring NEXT and Objective-C …
• Mikel Evins about the Lisp-based Newton OS. - A bit of history about an alternative Newton OS in Dylan. Those Newton guys where years ahead of their time!
• Monofur Font | dafont.com - My current favourite programming font. Monospaced with some nice ligatures and slanted characters to make it interesting to look at without getting in the way.
• Shared Library Search Paths -
• noelwelsh’s assembler at master - GitHub - This is an assembler for 32-bit Intel x86 code (IA-32) written in PLT Scheme.

I wrote about about why these posts are being generated here: Sharing Links.

Which I realise is *quite* the lofty goal, so let’s reel it in a bit. I’m making the rest of this year “Make Moblin Development Not Suck” time.

Right now we have a pretty poor offering for third party developers; as with many such projects we’ve been hugely user-oriented up until now.
Our little team is working on making this right, we’ll hopefully have some useful stuff to show soon.

For myself, I’m trying to address the Moblin Foundations and make them more approachable. We have some sweet libraries in there but they are largely a black box to developers due to low Gtk-doc annotation coverage outside of the core UI toolkits (Nbtk & Clutter).

Over the coming weeks I aim to address this; I’m looking through the code and trying to understand the APIs enough to get slightly better than skeletal Gtk-doc annotations added to our platform.

As a bonus prize with Gtk-doc annotations come almost free introspection metadata!

So, you’ve read this far and for that I thank you, as a reward here are the two lessons I’ve learnt so far this week:

1. Makefiles care about tabs, you can’t align with spaces
2. Building GObject-Introspection metadata outside of the standard, system, prefix. Tweak your environment!
   GI_TYPELIB_PATH is where the gobject-introspection module looks for typelib files (perhaps add /usr/local/lib/girepository-1.0) and the giscanner looks in XDG_DATA_DIRS for gir-1.0 directories chock full of .gir files (maybe add /usr/local/share)

Because of this I’ve decided that I’m going to keep a set of notes against ConnMan git for budding UI developers.

These notes will sit on this here website and be updated each time there’s a change in the ConnMan interfaces.
Once the interfaces stabilise for ConnMan 1.0 I’ll write the information up nicely and post on the ConnMan website or some such.

See the first draft here: Writing connectivity UIs for ConnMan.

Hung out with a bunch of cool people, attended some great talks, absorbed the general atmosphere and got inspired to do some hacking.

Guadec hacks:

• Started working on a ConnMan plugin to implement the shared connectivity namespace I proposed in my talk. This needs some more work and may involve me getting more familiar with ConnMan internals so it’s not ready yet.
• GUPnP context manager for ConnMan. GUpNP has NetworkManager and Unix back-ends for creating context manager objects so I sat down and hammered a ConnMan implementation too.I didn’t have enough network to pull all of the dependencies to build it so I’m going to do that in the office this week and have Ross look over and hopefully merge it before next week end.
• Of course, I did some carrick and gconnman hacking and even more meta-hacking (discussing features and bugs with others and helping people get familiar with the code).

While I was away Thomas made the Moblin Gtk theme rock even harder and I merged some of Jussi’s awesome visual enhancements to Carrick such that when I updated my netbook I got this piece of prettiness:

The software I’ve been working on for the last few months is starting to look really quite nice and it’s even getting to a stage where it’s mostly usable. Great stuff!

Guadec has inspired me to work on a few more things which I hope to be blogging about soon. My current #1 home-time-hacking priority is gconnman; I want to add GObject introspection annotations and think I’ve worked out a way to make it slightly more efficient.

More on both these things soon as we all decided at Guadec that we need to blog more about the neat stuff we’re working on with Moblin. It truly is the nuts.

Of course there where various questions about NetworkManager; “Why aren’t you using it”, “Why don’t you switch to it now 0.7 is released”, etc. Questions I was entirely unable to answer, various hand waving, etc.

On the whole the talk was productive though, I proposed a shared D-Bus API for common connectivity queries and after the talk Darren Kenny of Sun Microsystems told me he was interested in implementing this for the OpenSolaris connectivity daemon NWAM. Win!
I’ll be working on a ConnMan plugin and getting in touch with Darren soon.

I’ve boldly thrown up a talks page where you can get at my slides.

GCDS is really great so far. Saw a good talk by Will Thomson on “Profiling and Optimising D-Bus APIs”. His D-Bus profiling tool, Bustle, looks great and I’m going to share it with the ConnMan team.

Back to the talks.