Styhead support #231
Styhead support #231
Conversation
|
Building was blocked again. Have restarted the container |
|
It seems to have errored https://github.com/DynamicDevices/meta-mono/actions/runs/11159243181/job/31017239139?pr=231 |
…m WORKDIR.
* Using S = ${WORKDIR} is no longer supported.
* UNPACKDIR is new contruct for do_unpack things in latest master we should be using that instead of WORKDIR for referencing those files.
* We don't know yet what changes will be needed to stay compatible with final styhead, but we already know that the last changes for UNPACKDIR aren't compatible with scarthgap, nanbield or others.
https://lists.openembedded.org/g/openembedded-architecture/message/2007
https://docs.yoctoproject.org/dev/ref-manual/variables.html?highlight=unpackdir#term-UNPACKDIR
Signed-off-by: alperak <[email protected]>
Signed-off-by: Zoltán Böszörményi <[email protected]>
Using ${PYTHON_PN} to switch between python and python3 is
a thing of the past.
Signed-off-by: Zoltán Böszörményi <[email protected]>
Otherwise the python module recipes in meta-mono fail to find their dependencies. Signed-off-by: Zoltán Böszörményi <[email protected]>
*** Experimental change. *** Both GNU make and ninja support options: * -j N (number of parallel jobs) and * -l N (start a new job only if load is under N) Tune these so the CI runs may be a little faster. Signed-off-by: Zoltán Böszörményi <[email protected]>
Signed-off-by: Zoltán Böszörményi <[email protected]>
zboszor
force-pushed
the
styhead-support
branch
from
864c6e8 to
817ea46
Compare
I rebased over your |
This fixes the error: ERROR: libgdiplus-6.0.5-r0 do_patch: QA Issue: Missing Upstream-Status in patch /__w/meta-mono/meta-mono/styhead/meta-mono/recipes-mono/libgdiplus/libgdiplus-6.0.5/0001-fix-cross-compile.patch Please add according to https://docs.yoctoproject.org/contributor-guide/recipe-style-guide.html#patch-upstream-status . [patch-status] Signed-off-by: Zoltán Böszörményi <[email protected]>
Signed-off-by: Zoltán Böszörményi <[email protected]>
Signed-off-by: Zoltán Böszörményi <[email protected]>
Signed-off-by: Zoltán Böszörményi <[email protected]>
These are already fixed. Let's see if the last commit goes through. |
|
The errors are:
I will try to fix these the next week. |
Added two patches: * one to fix implicit cast issues by using explicit casts * one to fix warnings in the test application TestRange.cs Signed-off-by: Zoltán Böszörményi <[email protected]>
|
Either we can ignore the errors with a series of |
I'm in your hands here @zboszor. On the one hand the INSANE_SKIP seems like a bit of a hack (although I recognise this just means it would be working the way it always has worked). On the other hand the other option feels like it might be a fair bit of work. I'll go with whatever you suggest? |
I am already working on the second option using TL;DR: Then I was getting a differerent compiler error when building Now I am getting I am getting somewhere with very small changes for now. |
Signed-off-by: Zoltán Böszörményi <[email protected]>
|
I had to take a turn in the approach, which indeed took more time to find and fix everything. It turned out that passing Instead, I had to patch the offending It works on my side with x86-64 host and target, let's see what CI has to say. |
|
Fingers crossed! |
|
There are no more failures. Especially this is what we were waiting for: |
|
Great news! I am keeping an eye on the builds here too. I assume the rest will all go through and then I'll merge in. Thanks for your efforts as ever @zboszor ! |
|
NB. Once that goes into master I'll also make a new styhead branch based on that. |
|
As a side note, this last patch may be backported to the scarthgap branch because this way the mono build becomes reproducible. Probably some of the other changes, except the ones with the WORKDIR -> UNPACKDIR transition that are styhead+ only. |
|
Another side note: since we are not testing anything but the preferred version (6.12.0.182 at the moment) it's very likely that I broke older recipes. Don't we want to clean them up eventually? |
|
Wow. GCC 14 does not like 32-bit arm. https://github.com/DynamicDevices/meta-mono/actions/runs/11232177725/job/31223236463?pr=231#logs |
GCC 14.2 does not compile for arm, but it does for arm64. Signed-off-by: Zoltán Böszörményi <[email protected]>
That's fantastic. Love it. I have limited time with all the stuff going on here but would love to get on top of fully reproducible Yocto builds for our devices
I am sure we do. All I can say is that we're all giving our time here and whilst I am relying on the CI to flag up big problems I think my current thinking is that things will inevitably break somewhere and if it's a problem somebody will flag it and then we can figure out what's happened and ideally add a new test to the CI. I'm quite keen on adding new CI tests as we go along to try to improve regression testing.
Yikes. Well I guess the positive is at least we're seeing this now! |
|
CI is green now. |
|
So I created a discussion about dropping support for ARM32. What do we think? |
I am not very familiar with ARM32 but it is possible that other subarchitectures have no problem, it's just the FWIW, the ARM32 CI test can be easily revived when the GCC 14 build is fixed in OE-Core. |
|
Merged in then. Thanks ! |
Backports from #231 for scarthgap
This is a rebase of #225 over current master.
In the meantime, the styhead branches were created in upstream Yocto and styhead already diverged from current master leading to Yocto 5.2 "Walnascar".
I added the necessary CI change to use the styhead branch from poky. Not sure if this also uses (the currently non-existing) styhead branch from meta-mono, or as intended, its master branch.