-
Notifications
You must be signed in to change notification settings - Fork 533
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
R8 integration #1489
R8 integration #1489
Conversation
General guideline for sanity (which we're breaking in a couple places, and should fix, but let's not make it worse):
As such, |
We distribute mono, libzip and LibzipSharp, and they are under build-tools. |
dotnet#1489 (comment) complains about the additions of new sources to build-tools is wrong because they are not "build-tools", mentioning that the existing sources are wrong. Wait, then, why not fix it right now. Therefore, fixing this. `mono-runtimes`, `libzip` and `libzip-windows` are moved to src.
dotnet#1489 (comment) complains that the additions of new sources to build-tools is wrong because they are not "build-tools", mentioning that the existing sources are wrong. Wait, then, why not fix it right now. Therefore, fixing this. `mono-runtimes`, `libzip` and `libzip-windows` are moved to src.
#1495 has been merged. For your next trick you should move |
This PR also needs to add |
edaf831
to
93dfb46
Compare
93dfb46
to
1b19111
Compare
I don't quite understand. Why don't you do it by yourself? Just do it. I will review your PR. |
9c3da1a
to
272e22c
Compare
There is no documentation about those .tpnitems file format (MSBuild Item description) and I'm not an ESPer. What do people specify values at |
272e22c
to
9cf1633
Compare
I should write up some documentation. In the meantime, try to copy what already exists. :-D
Generally, it would be "organization-name/repo-name" like github.com convention, e.g. However, there are no requirements on the value; it's just a short (non-empty) string that is used as a "section header" within the For this PR, I'd suggest values of |
Please update |
9cf1633
to
5d8137d
Compare
@eno: the macOS+xbuild PR Build fails with:
Additionally, that error message is the only "match" for Further reading the log, it appears that the Offhand, I'm not immediately sure why it's not being built. I would suggest that you update |
5d8137d
to
d228827
Compare
d228827
to
7b2cd6a
Compare
Context: https://jenkins.mono-project.com/job/xamarin-android-pr-builder/2963/testReport/ A build for PR #1489 ran 1110 tests; it *should* have run over 35,000. Where are the missing tests? What's missing are the `.apk` on-device tests; BCL tests make up *most* of our expected test count. Analysis of the build log shows that the on-device tests *are* running, with test results being downloaded: …/android-toolchain/sdk/platform-tools/adb -s emulator-5570 pull "/data/data/Xamarin.Android.Bcl_Tests/files/.__override__/TestResults.xUnit.xml" "…/xamarin-android/tests/../bin/TestDebug/TestResult-Xamarin.Android.Bcl_Tests.xunit.xml" ... /data/data/Xamarin.Android.Bcl_Tests/files/.__override__/TestResults.xUnit.xml: 1 file pulled. 53.5 MB/s (3273412 bytes in 0.058s) but those results don't appear on the `testReport` URL. The problem is that the `RenameApkTestCases` target is never being executed, and the `RenameApkTestCases` is responsible for copying the `TestResult*` files where Jenkins looks for them: Done building target "ReleaseAndroidTarget" in project "RunApkTests.targets" -- FAILED.: The `RenameApkTestCases` target executes *after* `ReleaseAndroidTarget`, but because `ReleaseAndroidTarget` failed, the `RenameApkTestCases` target isn't executed. Meanwhile, `ReleaseAndroidTarget` reports an error to ensure that errors aren't overlooked; see commit 3b893cd. Allow the `RenameApkTestCases` target to execute by adding a new `ReportComponentFailures` target. The `ReleaseAndroidTarget` will no longer report errors, allowing the `RenameApkTestCases` target to execute, while if any tests *do* fail, the `ReportComponentFailures` target will generate an appropriate error. Furthermore, update all the `<MSBuild/>` invocations within `RunTests.targets` and add `ContinueOnError="ErrorAndContinue"` to all of them, to help ensure that we run all the tests we have, even if previous tests fail.
1e9a437
to
c4c0e9f
Compare
c4c0e9f
to
f18dbd4
Compare
build |
f18dbd4
to
5c8ba5b
Compare
build |
#1792 is still blocking further investigation. |
Locally |
5c8ba5b
to
e3862a3
Compare
build |
Why is this Azure problem triggered only here? I'm seeing other newer builds pass and I'm not happy to see this.
|
e3862a3
to
7ca3a12
Compare
sigh, *.sln merge failure. |
The latest build failure does not make sense to me at all:
This is Anyways I'm removing this entry. |
fc41d96
to
c2d2102
Compare
I'm really tired of these failing PR builders which keeps intermittently fail for irrelevant issues. Can this PLEASE be merged NOW? |
c2d2102
to
d5ddba9
Compare
After some investigation jonp found that Mono.Android-Tests failed (not even starting tests) due to bogus app build triggered by r8 change and the details are simply suppressed, therefore looking like failing for no reason. The build tasks likely have issues and should be fixed first. |
Development is blocked by https://xamarinhq.slack.com/files/U03CZ68AY/FBKAEBB0D/screenshot_from_2018-07-05_14-18-37.png |
depot_tools is required to build r8. On Windows it is however not used but a binary executable zip will be downloaded and used instead. (see https://commondatastorage.googleapis.com/chrome-infra-docs/flat/depot_tools/docs/html/depot_tools_tutorial.html#_setting_up for details) r8 build is not integrated in the primary Makefile and .sln yet. (r8 is not a "build tool" but we build msbuild tasks dependencies there too.)
See dotnet#1423 for details. multidex support is not done yet (needs to investigate what's expected there). Xamarin.Android.sln has a lot of changed lines because they had wrong ProjectTypeGuids (maybe regression from changes to .csproj from .*proj).
d5ddba9
to
21d9c27
Compare
Closing this one in favor of #2019, where I'll be taking this over. I'll keep @atsushieno's commits there intact, and hopefully we can use a merge commit so we keep that history. |
(testing CI builds; needs more work)