Skip to content

Conversation

@nagilson
Copy link
Member

We want to ensure that nothing breaks when we refactor code and that we have proper test driven development. #51349 started this but we needed to have the pipeline file pushed into the branch to make the pipeline. This actually fixes the pipeline and makes it work.

im not sure if I really like this - chose a fortress to represent 'guarding' the code by running tests.... is it really helpful? It might be if / when we have other test steps to differentiate and find the dnup test checks.
there is no concept of a global\ mutex prefix on unix which could be incorrect. On windows, this makes it so the mutex is not for the local user session but instead across all users. I'm not sure if this is needed but since we edit the system path, it might be. I will have to think about whether there will be issues with a higher-level admin user that has held onto a mutex and got stuck, and whether a lower level user could undo that.
10 seconds may fail - concurrently the tests could take a while. lets not fail the test because other tests are running.
OpenExisting may run into challenges with security policies changing - let's not try to navigate that when we can track it ourselves.
this would likely cause concurrency issues in tests if we just rely on the environment variable.
tests total runtime is 200 s on my machine - let's see if this helps. this would ideally be configurable because most users probably won't run multiple installs, though also they may have a much slower machine where this would cause concurrent requests to fail.
may refactor this later...
we probably shouldnt hard code debug amongst other things. Im sure theres a better approach but just want to get it green first.
This implements validation of installs using hostfxr apis to ensure the install actually works and not just that the manifest is tracking the installs correctly in e2e tests.

It also adds a test to show that we can do multiple installs in the same directory without failing.

It also improves the existing test logic to not assume a hard-coded debug value for the dnup process.
@nagilson nagilson marked this pull request as ready for review October 21, 2025 18:04
@nagilson nagilson requested a review from MiYanni as a code owner October 21, 2025 18:04
@nagilson
Copy link
Member Author

Just going to merge this so other code isn't blocked - will fix the hardcoded debug expectations / whitespace later, then add further validation and tests in another PR.

@nagilson nagilson removed the request for review from MiYanni October 21, 2025 18:05
@nagilson nagilson merged commit 5ca205c into dnup Oct 21, 2025
13 checks passed
@nagilson nagilson deleted the nagilson-dnup-pipeline-fixes branch October 21, 2025 18:25
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant