Re: [Patch-gnuradio] [PATCH 0/4] usrp: libusb-1.0

[Eric Blossom]

On Tue, Aug 18, 2009 at 01:11:51PM -0400, Thomas Tsou wrote:
> On Mon, Aug 17, 2009 at 9:08 PM, Eric Blossom <address@hidden> wrote:
> 
> >
> > Thanks Thomas!
> >
> > In the places where you removed the usb_strerror(), is there any way
> > to get additional information about why the calls failed?
> >
> > Did you happen to do any cpu utilization comparison between the 0.12
> > and 1.0 versions?  oprofile?
> >
> 
> libusb-1.0 functions return negative error codes rather then errno type
> reporting. I converted many error statements to reflect this. Still, the
> usb_strerror() calls were stripped out in a single pass, as there was no 1.0
> equivalent; so I'm sure that a few libusb return values remain to be
> checked.

OK.  Thanks for explaining.

> No utilization comparisons at the moment. This was basically an interesting
> side project that could be done rather quickly. I'll need to find some time
> to put oprofile on it. This weekend perhaps?

That would be great.


Getting it working with Tx and Rx simultaneously is important,
otherwise none of the transceiver apps (e.g. tunnel.py) will work.

The second thing to work out is how to cleanly support both 0.12 and
1.0 without a bunch of code duplication.  At least for a while, we'll
need to support both, selected at configure time based on what we find
on the system.  I've got some vague ideas on this, but don't have time
to spell them out right now.  Typedefs for the types that changed
between the old and new, but serve the same purpose, would be part of
the strategy.

Johnathan Corgan will be contacting you about some git related info
a bit later on.


Thanks again for working on this!

Eric