On 1/5/26 7:44 PM, Denys Dmytriyenko wrote:
There's a standard DEBUG_BUILD variable that can be reused for this. But...
It is a build-time condition, which means you end up with two separate builds
for release and debugging. Do you test both? Or how can you ensure testing
debug build also covers the release build properly?
The idea of only building release images and then installing test packages at
run-time for testing was to mitigate the above ambiguity...
I believe it is always best to test the same configuration we release, so I'd
like to avoid splitting the build. Instead simply looking to have a cleaner
split between images with testing packages and images without (all built with
the same configuration). Having the ti-test packages installed after by the
test infra itself is also a reasonable solution.
That said, I don't understand why it has to drive the run-time testing over
telnet or ssh? Isn't the early boot testing already done over serial? Can the
user-space Linux testing be done over serial as well?
This would be much better. As Ryan pointed out, even SSH doesn't save us
completely as we would need to still inject login info and such. Serial at
least requires some level of physical access to the HW before getting root.
Not perfect, but so much better than open root exposed over network.
Andrew
-=-=-=-=-=-=-=-=-=-=-=-
Links: You receive all messages sent to this group.
View/Reply Online (#16938):
https://lists.yoctoproject.org/g/meta-arago/message/16938
Mute This Topic: https://lists.yoctoproject.org/mt/117096038/21656
Group Owner: [email protected]
Unsubscribe: https://lists.yoctoproject.org/g/meta-arago/unsub
[[email protected]]
-=-=-=-=-=-=-=-=-=-=-=-
