make: fix export features#6690
Conversation
miri64
left a comment
There was a problem hiding this comment.
ACK & go, when Murdock is happy.
|
(needs rebase though) |
b9485c9 to
f0ff5a1
Compare
|
Rebased |
|
So the side effects in buildtest are resolved? |
|
needs further review, postponed |
Review dismissed because of @kaspar030's comment.
|
Good catch @smlng, thanks! |
smlng
left a comment
There was a problem hiding this comment.
ACK from my side, still I'd like to here @kaspar030 opinion on this before merge.
|
@lebrush I think you should squash (or drop) the merge commit - so the commit history in RIOT is clear from foreign merge commits |
412d9d2 to
ff76b60
Compare
|
There you go :-) sorry for the lack of responsiveness lately |
smlng
left a comment
There was a problem hiding this comment.
@kaspar030 is right, there is still a problem when running make buildtest. With this PR make buildtest builds tests/periph_rtc for all boards, though most of them do not have RTC. Using master, it correctly compiles only for boards with feature RTC.
|
see #5128 |
|
should wait for #7507 |
|
#7507 was merged, giving this PR a last round of testing/verification now... |
haukepetersen
left a comment
There was a problem hiding this comment.
PR works as expected (at least for all buildtests/info-boards-supported calls that I tried). Also the optional inclusion of the RTC shell command works as expected on nucleo-f411 (included) and arduino-due (not present).
ACK
|
@smlng ping |
|
all green -> go |
Fixes #2058
The current approach in the build system is to export all the variables required by the
make -C ....Another option would be to create a
Makefile.buildin theriotbuildfolder and import it inMakefile.base.