On Friday 26 September 2008 02:08:40 Eric B. wrote:
On Tue, Sep 16, 2008 at 05:47:56PM +0200, Stefan Brüns wrote:
[…]
This message didn’t have an attachment. Was there supposed to be one?
Actually, yes, but my fault.
FYI, Tom R., who’s been working on this stuff, has been traveling
for the last couple of weeks. Thanks for sending the patches. Either
he or I will take a look at them soon.
No worries.
Part of the strangeness that you may have seen in the original code,
is that there is class of daughterboards where the Tx and the Rx side
share resources such as a single LO (XCVR2450 and the not yet released
WBX-*). Either side needs to be able to access and control the shared
resource, which may be physically connected to the other side’s control
lines. I’m by no means sure that what’s in the tree is optimal by any
definition, but that’s part of what’s behind the design.
Yeah, the shared parts make a clean design challenging. Nevertheless the
biggest part of 2nd patch is IMHO still correct.
There is no point in setting a variable depending on the class and
calling a
dispatcher in the base class. So the only part which should be kept are
the “usrp_basic::common_*” methods, as these may be called from some
daughterboard classes keeping only an pointer to an usrp_basic instance
(more
or less).
Thanks again for your contributions!
What but be the best way to get any changes into trunk or a branch as
soon as
possible? First, there are some changes possible in Toms branch to get
it
more in line with trunk. Clearly there are other parts which are not in
trunk, but in Toms branch (eg the usrp_basic::db(…) method and related
stuff), which have to be merged into trunk. And there are some changes
which
should be applied to Toms branch to make the patch and the codebase as
small
as possible.
Stefan
–
Stefan Brüns / Bergstraße 21 / 52062 Aachen
mailto:lurch at gmx.li http://www.kawo1.rwth-aachen.de/~lurchi/
phone: +49 241 53809034 mobile: +49 151 50412019