* Test ci baseline change effect
* Skip parent hashes when baseline was modified
* Update head baseline variable
* Revert "Test ci baseline change effect"
This reverts commit a5e7a5c404eb400476555b92e51b3f5a87880300.
* Read raw data
* Update vcpkg-tool to 2022-03-24
* Hook up https://github.com/microsoft/vcpkg-tool/pull/345
* Hook up https://github.com/microsoft/vcpkg-tool/pull/442
* Update vcpkg-tool to 2022-03-25
* Analysis of failures.
* [Most recent nightly build failed](https://dev.azure.com/vcpkg/public/_build/results?buildId=69427)
* [Validation of this tool update failed](https://dev.azure.com/vcpkg/public/_build/results?buildId=69417)
## Common to both:
PASSING, REMOVE FROM FAIL LIST: chartdir:x64-windows (.\scripts\ci.baseline.txt)
PASSING, REMOVE FROM FAIL LIST: chartdir:x64-windows-static-md (.\scripts\ci.baseline.txt)
PASSING, REMOVE FROM FAIL LIST: chartdir:x86-windows (.\scripts\ci.baseline.txt)
Probably fixed by https://github.com/microsoft/vcpkg/pull/23701
PASSING, REMOVE FROM FAIL LIST: gmp:x64-uwp (.\scripts\ci.baseline.txt)
PASSING, REMOVE FROM FAIL LIST: gmp:x64-windows-static-md (.\scripts\ci.baseline.txt)
Probably fixed by https://github.com/microsoft/vcpkg/pull/23466 ?
REGRESSION: colmap:x64-windows-static-md failed with BUILD_FAILED. If expected, add colmap:x64-windows-static-md=fail to C:\a\2\s\scripts\azure-pipelines/../ci.baseline.txt.
I don't know exactly what changed. I observe that
* this thing depends on a *lot* of stuff
* on March 14 we didn't even attempt to build this
* the x64-windows ones are already in the baseline
so I skipped it.
REGRESSION: qtdeclarative:x64-windows. If expected, add qtdeclarative:x64-windows=fail to .\scripts\ci.baseline.txt.
REGRESSION: qtdeclarative:x64-windows failed with BUILD_FAILED. If expected, add qtdeclarative:x64-windows=fail to C:\a\1\s\scripts\azure-pipelines/../ci.baseline.txt.
REGRESSION: qtdeclarative:x64-windows failed with BUILD_FAILED. If expected, add qtdeclarative:x64-windows=fail to C:\a\2\s\scripts\azure-pipelines/../ci.baseline.txt.
REGRESSION: qtdeclarative:x64-windows failed with BUILD_FAILED. If expected, add qtdeclarative:x64-windows=fail to C:\a\2\s\scripts\azure-pipelines/../ci.baseline.txt.
REGRESSION: qtdeclarative:x64-windows failed with BUILD_FAILED. If expected, add qtdeclarative:x64-windows=fail to C:\a\2\s\scripts\azure-pipelines/../ci.baseline.txt.
REGRESSION: qtdeclarative:x64-windows failed with BUILD_FAILED. If expected, add qtdeclarative:x64-windows=fail to C:\a\2\s\scripts\azure-pipelines/../ci.baseline.txt.
REGRESSION: qtdeclarative:x64-windows failed with BUILD_FAILED. If expected, add qtdeclarative:x64-windows=fail to C:\a\2\s\scripts\azure-pipelines/../ci.baseline.txt.
REGRESSION: qtdeclarative:x64-windows failed with BUILD_FAILED. If expected, add qtdeclarative:x64-windows=fail to C:\a\2\s\scripts\azure-pipelines/../ci.baseline.txt.
This is a reporting change: The new world order also includes host build failures which is why it's duplicated.
See also https://github.com/microsoft/vcpkg/issues/23714
See also https://github.com/microsoft/vcpkg/issues/23490
I'm nervous about baslining this because it seems most of the qt world is built on top of this port
I filed https://github.com/microsoft/vcpkg/issues/23824 about this and @Neumann-A indicated this should be fixed by https://github.com/microsoft/vcpkg/pull/23755
REGRESSION: nettle:x64-uwp. If expected, add nettle:x64-uwp=fail to .\scripts\ci.baseline.txt.
REGRESSION: nettle:x64-windows-static-md. If expected, add nettle:x64-windows-static-md=fail to .\scripts\ci.baseline.txt.
REGRESSION: nettle:x64-uwp failed with BUILD_FAILED. If expected, add nettle:x64-uwp=fail to C:\a\2\s\scripts\azure-pipelines/../ci.baseline.txt.
REGRESSION: nettle:x64-windows-static-md failed with POST_BUILD_CHECKS_FAILED. If expected, add nettle:x64-windows-static-md=fail to C:\a\2\s\scripts\azure-pipelines/../ci.baseline.txt.
Didn't analyze, probably fixed by https://github.com/microsoft/vcpkg/pull/23519 ?
REGRESSION: libgpg-error:x64-uwp. If expected, add libgpg-error:x64-uwp=fail to .\scripts\ci.baseline.txt.
REGRESSION: libgpg-error:x64-uwp failed with BUILD_FAILED. If expected, add libgpg-error:x64-uwp=fail to C:\a\2\s\scripts\azure-pipelines/../ci.baseline.txt.
This was broken by VS2022 update:
```
C:\Program Files\Microsoft Visual Studio\2022\Enterprise\MSBuild\Microsoft\VisualStudio\v17.0\AppxPackage\Microsoft.AppXPackage.Targets(892,25): error MSB4086: A numeric comparison was attempted on "$(TargetPlatformMinVersion)" that evaluates to "" instead of a number, in condition "'$(TargetPlatformMinVersion)' >= '10.0.17200.0'". [C:\Dev\vcpkg\buildtrees\libgpg-error\x64-uwp-rel\error-1.42-2324ddbc71.clean\SMP\libgpg-error_winrt.vcxproj]
```
REGRESSION: libmikmod:x64-osx. If expected, add libmikmod:x64-osx=fail to .\scripts\ci.baseline.txt.
REGRESSION: libmikmod:x64-osx failed with BUILD_FAILED. If expected, add libmikmod:x64-osx=fail to /Users/vagrant/Data/work/2/s/scripts/azure-pipelines/../ci.baseline.txt.
Broken between [2022-03-16](https://dev.azure.com/vcpkg/public/_build/results?buildId=68947) and [2022-03-18](https://dev.azure.com/vcpkg/public/_build/results?buildId=69051). Unfortunately I don't see obvious reasons why. Nothing else depends on this and nobody has noticed in 2 weeks, so I'm baslining it for now. (Will investigate shortly...)
## Only broken in tool update:
REGRESSION: mesa:x64-windows failed with BUILD_FAILED. If expected, add mesa:x64-windows=fail to C:\a\2\s\scripts\azure-pipelines/../ci.baseline.txt.
```
-- Downloading https://gitlab.freedesktop.org/mesa/mesa/-/archive/mesa-21.2.5/mesa-mesa-21.2.5.tar.gz -> mesa-mesa-mesa-21.2.5-1.tar.gz...
-- Extracting source /Users/vagrant/Data/downloads/mesa-mesa-mesa-21.2.5-1.tar.gz
-- Applying patch swravx512-post-static-link.patch
-- Applying patch swr-msvc-2.patch
-- Applying patch swr-llvm13.patch
-- Applying patch radv-msvc-llvm13-2.patch
-- Applying patch d3d10sw.patch
-- Using source at /Users/vagrant/Data/buildtrees/mesa/src/esa-21.2.5-2df234d2b1.clean
Traceback (most recent call last):
File "<string>", line 1, in <module>
ModuleNotFoundError: No module named 'mako'
CMake Error at ports/mesa/portfile.cmake:85 (message):
Python package 'mako' needs to be installed for port 'mesa'.
Complete list of required python packages: setuptools;mako
Call Stack (most recent call first):
ports/mesa/portfile.cmake:91 (vcpkg_get_python_package)
scripts/ports.cmake:145 (include)
```
Looks like this is being tracked by https://github.com/microsoft/vcpkg/pull/23089 ; perhaps that we don't have as aggressive a recycling strategy for macos boxes as we do for the others has let different machines give different results?
## Only broken without tool update:
REGRESSION: chromium-base:x64-osx. If expected, add chromium-base:x64-osx=fail to .\scripts\ci.baseline.txt.
This one has been constantly flaky; I baselined it.
REGRESSION: libxml2:x64-osx. If expected, add libxml2:x64-osx=fail to .\scripts\ci.baseline.txt.
This port uses vcpkg_from_git and the upstream server was down during the build.
* Restore chartdir to the baseline, I thought https://github.com/microsoft/vcpkg/pull/23732 had been merged.
* Trigger some test ports from vcpkg.cmake changes
* Bump test port version
* Revert "Bump test port version"
* Use old toolchain version for parent hashes
* Add ABI hashing note to vcpkg.cmake
* Test vcpkg.cmake with cmake instead vpckg-ci-paraview
* Update vcpkg-tool to 2022-03-08.
* Back off required git version to something actually in the build lab.
* Remove hiding of the output of --parent-hashes.
* Update 7zip reference as indicated in https://github.com/microsoft/vcpkg-tool/pull/413
* Add triplet to name of main CI task
* Fix osx bootstrap task name
* Hide output from switch to parent hash
* Bump test port version
* Hardcode osx+linux triplet names
* Improve logging order
* Add error handling to each tool download
* Revert "Bump test port version"
* Don't prefetch 7zip
* Rewrite to single invovation of 'vcpkg ci'
* Split caching args from common args
* Use parent hashes for pull request CI builds
* Completely switch to parent revision state and back
* [macos ci] move out tensorflow
* rename ExtraPorts and CorePorts
Co-authored-by: nicole mazzuca <mazzucan@outlook.com>
Co-authored-by: Billy Robert O'Neal III <bion@microsoft.com>
* Decouple the storage account used for binary caching from the images.
We are being forced to change infrastructure somewhat more frequently than we would like, so change the binary caches to use a shared storage account like we already use for asset caching.
* blob blob blob
* Add quotes
* Avoid name stomping.
* [vcpkg] Update vcpkg tool version 2021-05-05
* [vcpkg] Hotfix bug in Microsoft/vcpkg-tool#34
* [vcpkg-ci] Ignore results from non-target triplets
* [protobuf] Fix regression introduced in #16904
* [vcpkg] Fix bug where 'staticcrt' was ignored
Co-authored-by: Robert Schumacher <ras0219@outlook.com>
* Add meson from https://github.com/microsoft/vcpkg/pull/12860/
* Add autoconf-archive from https://github.com/microsoft/vcpkg/pull/13081/
* Add kf5windowsystem libs from https://github.com/microsoft/vcpkg/pull/13467/
* Open the FTP and SFTP ports from https://github.com/microsoft/vcpkg/pull/14412/
* Add libxcb-util0-dev from https://github.com/microsoft/vcpkg/pull/14678/
* Add libasound2-dev from https://github.com/microsoft/vcpkg/pull/14774
* Remove no longer necessary apt-mark calls.
* Update nasm on Linux.
* Fix longstanding bug where Storage was publicly accessible and change to generate SAS token rather than File Share
* Delete no longer necessary azure storage firewall rules.
* Install the newer Windows SDK with the VS installer instead of manually.
* Install the VS2015 and VS2017 compilers.
* Update Powershell-Core to 7.1.0.
* Update source of WDK.
* Update pools.
* [opentracing] Repair arm64-windows failures caused by mojibake in `expected.hpp` and errors in opentracing-cpp's lint for arm64 where it thinks exceptions are disabled when they are enabled.
Fixes:
C:\Dev\vcpkg\buildtrees\opentracing\src\b67575dab0-0250653c81.clean\3rd_party\include\opentracing/expected/expected.hpp(1): warning C4828: The file contains a character starting at offset 0x4a77 that is illegal in the current source character set (codepage 65001).
Fixes:
D:\buildtrees\opentracing\src\b67575dab0-0250653c81.clean\include\opentracing/tracer.h:223:5: error: cannot use 'try' with exceptions disabled [clang-diagnostic-error]
try {
^
* [mmloader] Patch out overrides of CMAKE_C_FLAGS and CMAKE_CXX_FLAGS that inserted /WX.
Note that this port sets /GS-, possibly because it may be for authoring shellcode.
* Bump storage API version to 2020-04-08.
* [vcpkg] Add experimental x-azblob binary provider
* [vcpkg] Test azblob storage provider in CI
* [vcpkg] Address some CR comments from #13639
* [vcpkg] Fixup azure-pipelines
* [vcpkg] Fix regression where the downloaded package is purged before decompressing
* [vcpkg] Further refactor vcpkg::Downloads
* [vcpkg] Enable OSX for x-azblob testing
* [vcpkg] Reduce diff against master
* [vcpkg] Extract Downloads::details::split_uri_view
* [vcpkg] Address PR comments
* [vcpkg] Add testing and metrics for x-azblob
* [vcpkg] Add docs for x-azblob
This includes a note that it is currently experimental
* [vcpkg] Address CR comments
* [vcpkg] Revert pipeline changes except OSX to minimize disruption
Co-authored-by: Robert Schumacher <roschuma@microsoft.com>
Co-authored-by: Billy Robert O'Neal III <bion@microsoft.com>
* [ampl-asl] Separate port ampl-mp module asl to ampl-asl
* [ampl-mp] Update version, separate port ampl-asl and fix arm build
* update baseline
* [VMSS] Pre-install ampl-mp:x86-windows
* Update ports/ampl-asl/portfile.cmake
* [cmake] add port for qt testing
* only add cmcldeps on windows
* [cmake] cleanup and fix portfile for osx build
* [cmake] more cleanup
* move cmake to scripts/test_ports
* add the overlay scripts/test_ports to ci
This changes our PR builds to treat 'fail' in the ci.baseline.txt as 'skip' instead of using tombstones.
We currently have large numbers of spurious failures that get enshrined in PRs through no fault of a PR author, removing the tombstones concept will fix those by allowing the user to retry. This does mean we accept some risk of not detecting when a port is 'fixed', but that failure is reasonable for us to handle after we see it in CI, but that seems worth it given that it lets us get rid of the tombstone concept.
This also helps out the binary caching feature, because we don't have to figure out how to productize tombstones.
