Most of the apps in Fedora that use D-BUS and HAL have been ported to the new API’s. May I never have to go through that again. NetworkManager remains as Dan is still not decided if he wishes to fork or create a compat library so that development can still move forward for the RHEL 4 releases. If we fork, patches would have to be constantly backported. We already have an untested compat library for the D-BUS API so it is all about HAL right now.
Packages can be found at http://people.redhat.com/johnp/redhat/experimental/utopia/i386/
I will be sending patches upstream as soon as they have been tested, so sometime next week. Hopefully we can make a new release of D-BUS and HAL by that time. Anyone doing HAL and D-BUS work should start to target the new API’s unless you need to support older versions of the libraries.
Here are a list of packages that have been modified:
- gnome-vfs
- gnome-volume-manager
- nautilus-cd-burner
- eggcups (patch already upstream)
- hal (patch already upstream)
[read this post in:
ar de es fr it ja ko pt ru zh-CN ]
Comments Off
A lot has changed in the D-BUS world and I am in the middle of making sure the apps that use D-BUS change with it. It is actually a fun job other than being a bit repetitive. It has allowed me to get better acquainted with the D-BUS codebase and understand its little quirks. I finsihed the mother of all D-BUS apps today by porting HAL’s marshaling code to the new D-BUS 0.30 codebase, fixing some D-BUS code in the process. The new recursive type system is really nice and for the most part intuitve. I was able to fix some hacks that HAL used to transfer hashes of string arrays with the new system. Next step is to port all the apps that use HAL to the new HAL API.
I have some changes on local disk that implements the rest of the python bindings to support all of the new D-BUS types. Structs will map to tuples and arrays of dict entries to hashes. I’m still having trouble thinking how I am going to support deep recursion. Because some containers need to be opened with the complete signature of its children I may have to bite the bullet and traverse data structures twice. Once to get the signsture and then again to marshal the values. I’m not quite sure why the containers couldn’t be constructed without signatures and then when the container is closed, recursivly generate them. After I am done with that I should take a look at implementing intospection for Python D-BUS objects and also using introspection to decide how to marshal parameters for those more exotic types like uint* which Python has no real concept of.
–
J5
[read this post in:
ar de es fr it ja ko pt ru zh-CN ]
Comments Off
At the Intersection of Reason and Dream was finnished yesterday at Crazy Hair Production’s studio in Purchase, New York. It is a four song compilation with a fifth 30 second outro which may be made into a complete song at some point. Time was running short so I didn’t get to record the title track which is my newest and is as of yet unfinished. I think it came out pretty good, at least as good as any album that took only five hours to put down. Most songs take weeks or months to get right in the studio. I will be posting under a creative commons license after Diana finishes the cover art and Travis (the studio engineer) sends me the final mixdown. I’m also thinking of doing a dual session CD with the oggs and ogg vorbis windows codex on the second session if there is enough room.
I just got back from NY and was looking forward to spending a nice anti-social night by myself. I’m not a big football fan and after Saturday in the studio I was pretty much drained. But, my social side got the best of me so I will be hanging out with Brian Clark, David Zuethen and the gang to watch the game. That is why, after all, I decided to live close to Boston and why I am going to be moving even closer come spring. David, I’ll bring the Jack.
[read this post in:
ar de es fr it ja ko pt ru zh-CN ]
Comments Off