Part of https://github.com/Azure/azure-sdk-for-cpp/issues/1277, checking what types of warnings the CI emits.
Verified all SDK product `.cpp`, `.hpp`, `.txt`, and `.md` files, primarily focused on azure-core. They are all clean now. There are some exceptions that needs to be added for keyvault and storage, but they are false positives, so not a concern.
> `cspell lint --config .vscode/cspell.json *.hpp */**/*.hpp`
CSpell: Files checked: 188, Issues found: 0 in 0 files
> `cspell lint --config .vscode/cspell.json *.cpp */**/*.cpp`
CSpell: Files checked: 186, Issues found: 88 in 15 files (keyvault and storage false positives, or tests)
> `cspell lint --config .vscode/cspell.json *.md */**/*.md`
CSpell: Files checked: 45, Issues found: 5 in 2 files (eng/common)
> `cspell lint --config .vscode/cspell.json *.txt */**/*.txt`
CSpell: Files checked: 44, Issues found: 0 in 0 files
> `cspell lint --config .vscode/cspell.json *.* */**/*`
CSpell: Files checked: 646, Issues found: 328 in 69 files (most of these are in eng\docs\api\assets\style.css or eng/common)
Deprioritize and ignored the errors from the test files (including test resource json files), and `eng/common` since those need to be centrally fixed.
DevOps helper scripts were updated in azure-sdk repo
so syncing those changes into the tools repo so they
can be used prepare-release script.
Co-authored-by: Wes Haggard <Wes.Haggard@microsoft.com>
This will allow us to point our verify-link script at a raw
MD file in a github repo and have it parsed correctly for
links.
Co-authored-by: Wes Haggard <Wes.Haggard@microsoft.com>
Co-authored-by: Wes Haggard <weshaggard@users.noreply.github.com>
* Use linux for vcpkg release
* Force lf line endings
* Set config correctly
* Better core.eol syntax and sequencing
* Try using linux host which uses LF by default
* Revert "Try using linux host which uses LF by default"
This reverts commit 160eb22b647e851a50f9d1e7b157a1e385fe946f.
* Use vcpkg.exe to format the manifest
* Spell correctly
* Move format-manfiest
- Update link checking pipeline to enable caching
- Add loading cache file from a http endpoint
- Add retry logic when hitting 429
Fix verify link renaming scenarios
- Correctly error if link format is incorrect
- Handle case where the replace regex is not passed
- Improve the error logging so it is easier to
identify the broken links
- This removes the ci.yml filtering which means all projects can be discovered
- This eliminates the need for the yml parsing module for most common scenarios
- This does introduce some potential issues with duplicate package names for
languages like java. The only known case currently is azure-storage-blobs and
if there is ever a need to explicitly pick the correct one someone can pass
in the ServiceDirectory filter similar to "storage/azure*" for new and "storage/microsoft*"
for the older sdk. By default azure usually comes first so the new one gets discovered
which is likely all we need for now. If this turns out not to be true we might
need another way to disambiguate.
Co-authored-by: Wes Haggard <Wes.Haggard@microsoft.com>
Co-authored-by: Wes Haggard <weshaggard@users.noreply.github.com>
* Update vcpkg ports to use a manifest json file instead of a CONTROL file.
* Update vcpkg ports to use a manifest json file instead of a CONTROL file
for remaining packages.
* Add the new vcpkg manifet json files and also update the template
package.
* Update paths in the cmake and eng scripts to find the right file now
that CONTROL is gone.
* Handle empty files in Test-SampleMetadata
Relates to Azure/azure-sdk-for-java#20684
* Remove condition from verify-samples
* Use forward slash in path
* Fix directory typo
Co-authored-by: Heath Stewart <heaths@microsoft.com>
* Verify samples during CIs
Replaces Azure/azure-sdk-for-python#17999 for use in all language repos.
* Move exit to proper scope
Co-authored-by: Heath Stewart <heaths@microsoft.com>
* Remove passing -Mode Complete to deployment
When passing Complete it will remove any resources
already in the resource group that weren't part
of the current deployment. That removal breaks a
lot of assumptions, like multiple deployments when
testing things like smoke-tests or if you are reusing
an existing resource group. We don't want that to happen.
* Remove ServiceDirectory as required parameter for remove/update
When this was made mandatory it broke some usages which didn't
pass the value. Those usages don't need to pass it because
the pass required information in other ways so removing the
requirement for the parameter to be passed.
Co-authored-by: Wes Haggard <Wes.Haggard@microsoft.com>
Initializing the BaseName with a ServiceDirectory that contains
a "/", because it is a multiple level path, causes the BaseName
initialization code to fail because it doesn't support the
validation pattern.
In all the known cases we already pass the ResourceGroupName
explicitly so don't need to set the BaseName so we can
skip the initialization in those cases.
Co-authored-by: Wes Haggard <Wes.Haggard@microsoft.com>
* Use SubscriptionId throughout TestResources
Fixes#1454
* Resolve PR feedback
* Default DeleteAfterHours to 48 for SDK team
Also makes a few other adjustments for subscriptions, like restoring the previous one if available and another was specified.
* Resolve PR feedback
* Change deployment mode to Complete
Also fixes an issue where if the user opted not to deploy to the same resource group, the script would continue execution anyway.
* Use consistent aka links to satisfy link checker
Only need it for the new Update-TestResources.ps1 script, but I wanted them to look consistent.
Co-authored-by: Heath Stewart <heaths@microsoft.com>
* Fix issue where 2d index arrays with length 1 get flattened by powershell
* Pass NonSparseParameters to imported matrix generation
Co-authored-by: Ben Broderick Phillips <bebroder@microsoft.com>
* Add chagelogpath to Update-ChangeLog.ps1
* Make PackageName and ServiceDirectory not mandatory, test ChangelogPath
Co-authored-by: Chidozie Ononiwu <chononiw@microsoft.com>
* Handle null values when generating display names
* Fix issue processing matrices where the base matrix is only an import
Co-authored-by: Ben Broderick Phillips <bebroder@microsoft.com>
