read packages from packages.config

[notofug]

Attempt to solve #15

Might be too naive but seem to work OK for initial test, please have a look and tell me what you think.

[sensslen]

Unfortunately the code you propose does not work properly. We offer a flag that allows to also print transitive dependencies (see https://devblogs.microsoft.com/nuget/introducing-transitive-dependencies-in-visual-studio/ for more information). The issue is that transitive dependencies are not considered for packages specified in packages.config.

To verify this works correctly, please add at least one additional integration test that checks that transitive dependencies are indeed followed.

Also I would love if you could use as much tooling already available instead of implementing the xml parsing on your own.... --> See https://stackoverflow.com/questions/33035704/how-to-read-the-list-of-nuget-packages-in-packages-config-programatically for an example

[sensslen]

Thank you very much for your contribution. Unfortunately there are currently two issues I see. Would you mind fixing these?

There are formatting issues.....
These can be resolved very easily by running dotnet format [path to solution]

Also there is an error 'ManagementClass' is only supported on: 'windows'

[notofug]

There are formatting issues..... These can be resolved very easily by running dotnet format [path to solution]
Also there is an error 'ManagementClass' is only supported on: 'windows'

I tried to isolate the use of ManagementClass to windows-builds only, which makes sense since msbuild c++ projects are locked to windows anyway as far as I am aware of. In fact this part of the PR deals explicitly with supporting c++ - projects, since such projects are limited to using 'packages.config' - style if using nuget.
Please check if this change fixes the build issue

[sensslen]

Ok, so there are still test failures - can you please run the tests locally to fix all of them?

[notofug]

I had to move the file-reading to a wrapper-class in order to satisfy the 'architectural test', please check if that makes sense or we should attack it in another way? (Tests seem to pass locally now at least ;-))

[sensslen]

Looks good now

[sonarcloud]

Quality Gate passed

Kudos, no new issues were introduced!

0 New issues
0 Security Hotspots
No data about Coverage
0.0% Duplication on New Code

See analysis details on SonarCloud