[fprint] [PATCH] misc patches

[Bastien Nocera]

On Wed, 2008-05-14 at 16:14 +0100, Daniel Drake wrote:
> Bastien Nocera wrote:
> > 0001-Add-i18n-support.patch: Adds i18n support
> > 0001-Clean-up-object-creation-code.patch: Clean up the object creation
> > in manager.c
> > 0001-Use-GObject-provided-boilerplate.patch: same thing for device.c
> 
> Applied all of them. I think you may have misunderstood one part though.
> 
> /* FIXME some or all of this should probably be moved into FprintManager
>   * ctor. however i'm not sure how to raise errors from a constructor...
>   */
> if (!discovered_devs)
>      return NULL;
> 
> You removed the above. I just want to point out that discovered_devs 
> being NULL indicates something truly went wrong somewhere (e.g. couldn't 
> allocate memory, couldn't enumerate USB devices, etc). It does not 
> indicate zero devices found.

Ha, ok. But in any case, we should error out when getting the error from
_get_error(), not when creating the object.

> If zero devices were found you still get a list back, it's just empty 
> (only has the NULL terminator).

I thought that was it. I'll fix it up then.

> > My plan is to add:
> > 1) user (of the api) tracking
> > 2) automatically exit when the daemon is unused for a while (as startup
> > is very fast)
> > 3) make sure the list of devices is updated when fprintd is running
> 
> Yep, I think we're in agreement on that. The 3rd item might be a little 
> tricky, given that libusb itself isn't very good at exposing device 
> additions/removals (this is planned for v1.1). We'll see how you get on 
> anyway :)

Yeah, the first one won't be easy either, but at least it won't be a
problem with libusb :)

We have similar problems in gvfs with gphoto and openobex using libusb
to enumerate devices, it sucks, especially as we can already enumerate
devices using HAL.

Cheers