areusch edited a comment on pull request #6703:
URL: https://github.com/apache/incubator-tvm/pull/6703#issuecomment-715743438


   @manupa-arm
   
   > From a high-level the need for the Vagrant boxes arise from the need not 
being able to use serial COM ports of Windows via the docker image, right ? 
(ref : docker/for-win#1018).
   
   correct. however, it's still a bit messy to configure serial ports, because 
the Zephyr udev rules need to be installed in the docker host/guest VM. 
further, they need to be individually forwarded with -v, so you need to 
prediect the /dev/ttyACM* names.
   
   > Out of curiosity, have you thought about spinning a docker up , possibly 
extending ci_qemu, to include the dependencies of Zehpyr and use it inside the 
Vagrant (perhaps via Docker Provisioner ? ) -- therefore the Vagrant will 
provide just the thing the docker is missing which is the serial COM port 
exposure (in Windows) inside the box.
   
   due to the aforementioned messiness, it seems better to just stick with one 
virtualization layer. 
   
   @manupa-arm @u99127 
   I've split out the `-mfpu` change, I agree that's fairly separate from this 
one.
   
   now with respect to the rest of the PR--I see that we could theoretically 
split it, but I don't know it gets us much. because much of this is python code 
that deals with talking to real hardware, it's not simple to develop and 
maintain unit tests for it. we can certainly add some, but the real utility is 
in an end-to-end test. that's what this PR adds--the end to end test and the 
minimum infrastructure needed to support it. further, it proves the end-to-end 
test is, at least at the user script level, so similar to the QEMU test ran in 
the CI, that it can be parameterized.


----------------------------------------------------------------
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.

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


Reply via email to