Re: State of the Pilot

[David Malcolm <dmalcolm redhat com>]

On Mon, 2005-03-28 at 15:37 -0500, JP Rosevear wrote:
> There have recently been several mails about what's happening with
> gnome-pilot development lately.  Basically as Dave Malcolm noted, its a
> time issue for me, althought I'm attempting to do better.

Thanks for this email, various random comments inline...

> 
> Several people have asked about the bonobo daemon branch, its a major
> overhaul I started months and months ago.  
> 
> It accomplishes several technical things:
> 
> 1. Removes GOB usage
Excellent!

> 2. Switches to gconf from gnome-config, and no longer requires the
> daemon to catch a signal it order to know to re-load config
> 3. Cleans up the architecture, each sync type (usb/serial/irda) is a
> class that implments the virtual class gnome-pilot-device
> 4. Uses bonobo object instead of dealing with all the corba cruft itself
> 5. Do gnome-pilot-client notifications via bonobo-listener/event-source
> 6. Simplifies conduit creation in some cases
> 7. Improved logging/error messaging
> 8. Probably some other stuff I've now forgotten

The UI could probably do with some love at some point as well, IIRC much
of the code predates the HIG.

> 
> As well as several usuability things:
> 1. Makes it easier for users by allowing the user to create a new
> "profile" during a sync.  ie the user can just put in a device press
> hotsync and set up the device without pre-configuring it
> 2. Simplifies the cradle setup (more windows like - you just check off
> "serial" or "usb" or whatever and enter the port - hal/dbus can
> hopefully can the usb port bit)
With HAL it should be possible to entirely automate USB device handling.
There are some questions as to where PDA handling should take place, my
vote is to push it down the stack into HAL as much as possible.

> 
> Its missing porting of conduits other than file/backup and some polish.
> Ideally I'd like new development to take place here.
Do you have a full list?  Any idea how much work would be needed to get
this into a releasable state?

> 
> So for 2.0.x:
> 
> I started going through patches in bugzilla this past weekend.  There
> are also patches on the list that need review.  Regardless of how far I
> get, I'll probably do a release in time for GNOME 2.10.1.
> 
> I'm aware of the HAL patches and i did briefly look at one a couple of
> months ago and I remember being unsatisfied, but I don't remember being
> why, so clearly I need to look again and give feedback.  It also looks
> like for several of the OS5 devices we are going to have to move to
> pilot-link 0.12.  I need to go back and look at DM's and Dave
> Desrosier's patches for this.
BTW there was a posting about PDAs to the HAL list today.

pilot-link-0.12 has a few API changes; buffers are handled explicitly by
pilot-link, so this requires lots of little changes in all of the
conduits.  We could in theory isolate this with a wrapper API, so that
the code can be used against both pilot-link-0.11 and 0.12.  Any
thoughts on how useful this would be?  Is it a goal to get
gnome-pilot-2.0.* to work against p-l-0.12, or are you thinking of doing
this for the bonobo-daemon-branch ?

> 
> Some things I'd like some help with:
> 1. Mark old 0.1.xx bugs as obsolete (except for feature requests)
> 2. Review all other bugs for duplicates
> 3. Debugging evolution pilot related issues - people with source
> installs of evolution and gnome-pilot preferrably with repeatable issues
> who can use gdb and possibly valgrind.
> 4. Testing of p-l 0.12 patches with OS5 devices that have issues
> currently (I believe a lot of these revolve around backing up files).

What's the preferred route for submitting patches?  This mailing list,
or as attachments to bugzilla?

Do we have an IRC channel?

Dave