acassis commented on PR #17363:
URL: https://github.com/apache/nuttx/pull/17363#issuecomment-3567108723

   > > > I think char device sensor still important on NuttX because for boards 
with low resource adding uORB will be too much. This is the same relation with 
CAN char device and SocketCAN, both implementations should exist to live 
together peacefully.
   > > 
   > > 
   > > Honestly I disagree. The requirement going forward for all sensor 
drivers was that they must support uORB, so we have a few drivers already that 
are uORB capable but do not have a legacy-style version, since most people do 
not want to do double the work. We definitely can't require people to submit 
both styles, so I don't think we should be supporting additional legacy drivers 
unless there's a use-case for them. I think supporting both versions is 
confusing to users and makes maintenance difficult. I also don't think uORB is 
that resource-intensive besides the floating point requirement, but it was 
decided when uORB became a requirement that the `float` dependency was fine.
   > > If the community still wants to support legacy sensor drivers in 
addition to uORB that's fine, I will yield, but the drivers will be perpetually 
out of sync which I think is a problem. Regardless, I still think the 
implementation in this PR needs changes if both styles are to be kept, since it 
currently requires shared header and C files that expose private parts of the 
API which are not exposed in any other sensor drivers on NuttX.
   > 
   > Yes, I agree that it could generate some drivers only supporting one or 
other implementation. Same also happens with CAN. It should be nice if uORB was 
just a wrapper around the char device, this way we could avoid this issue. But 
I don't know if it is possible. Maybe @xiaoxiang781216 or @raiden00pl has some 
suggestion how we could fix it to have sensors support on low end 
microcontrollers.
   
   


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]

Reply via email to