On 01/05/2010 11:21 AM, Michael Buesch wrote: > > Well, just do the thing that makes most sense. > In general we all agree that we should _not_ implement crap, just because > broadcom does so, if we can do better. So, in this case, if we can do a > subfunction call and that function does make sense, we should do so for the > sake of readability (I didn't look into detail, though.). > Same goes for algorithms and stuff. If we realize that we can do better, do > _not_ > implement Broadcrap and instead implement a better version.
I agree with Michael that we do not need to follow Broadcrap (nice phrase); however, as we usually have no idea of what is going on inside the chip, we do need to read/write exactly the same registers as they do without skipping any or touching any others. For the record, the specs will usually follow the Broadcrap way. I did come upon one exception recently. After quite a bit of decision making that did not touch any registers nor change any global variables, they suddenly tested for 2.4 GHz mode and exited if found. I moved that test to the beginning of the routine. Larry _______________________________________________ Bcm43xx-dev mailing list Bcm43xx-dev@lists.berlios.de https://lists.berlios.de/mailman/listinfo/bcm43xx-dev