Supporting line wise yanks, including paste and undo.

[springcomp]

This is my take at #333.

I thought best to create a ViRegister class to replace the current _clipboard String because I wanted to support pasting "line wise" irrespective of where the cursor is at when pasting occurs. I created an event that gets raised each time the buffer is about to change as part of a paste operation, to hook into undo/redo without too much coupling.

Please, let me know your thoughts and suggestions for improvements.

[springcomp]

I updated this pull request via force push using the accepted answer from this question.

[msftclas]

All CLA requirements met.

[lzybkr]

Sorry it took forever for me to take a look
I have some general comments that need to be addressed and I'll try to take a closer look then, or maybe @daxian-dbw will beat me to it.

[springcomp]

Sorry, I force-pushed again because there was structural changes to address your general comments.
You can now take a closer look.

From now on I will address your feedback with additional commits.

[springcomp]

This PR still shows changes requested even though I have carried out the changes.
@lzybkr can you please re-review?

[lzybkr]

It looks like not all of the new tests were run in CI - presumably because test discovery isn't searching Microsoft.PowerShell.PSReadLine2.dll.

I'd like @daxian-dbw to approve including tests like this before merging.

[springcomp]

The latest commit adds discovery and execution of thoses unit-tests.

For this purpose, the main module is compiled with an alternate set of options and to a target assembly named Microsoft.PowerShell.PSReadLine2.UnitTests.dll so that it does not conflict with the release assembly.

@daxian-dbw, are you OK with these changes?

[daxian-dbw]

Can we move the test code to PSReadLine.Tests (under PSReadLine\test)? We can make Microsoft.PowerShell.PSReadLine2 internal visible to PSReadLine.Tests.

[springcomp]

I thought about doing this but was deterred as per @lzybkr’s comment.

I would be happy to go ahead, but I do not know how to handle InternalsVisibleTo and assembly delay-signing?

For this to work, there would need to be a strong-name snk file in the repository for signing both assemblies. The way I handled this in the past was for the owner of the repository to include an encrypted version of the release strong-name snk file in the repository, and use AppVeyor secure strings to decrypt the file during the build.

Would that work for you @daxian-dbw?

[daxian-dbw]

I don't think InternalsVisibleTo requires the involved assemblies to be strong-name signed. Quoted the following from the doc of this attribute

If both assemblies are unsigned, the assemblyName argument consists of the name of the friend assembly, specified without a directory path or file name extension.

I think you can just add [assembly: InternalsVisibleTo("PSReadLine.Tests")] to PSReadLine\PSReadLine\AssemblyInfo.cs and it should work.

[springcomp]

I can’t believe I just learned about that fact today 😱😱!

[daxian-dbw]

Thanks @springcomp!
I made a minor change to move the InternalsVisibleTo declaration to AssemblyInfo.cs, where other assembly scope attributes are put. That change breaks the build and it seems that file gets excluded somehow. I will fix it.

[daxian-dbw]

Build fixed. @springcomp Could you please review my changes? Thanks!

[springcomp]

Yes, that is the reason why I created a separate file but this solution is certainly better.

[springcomp]

With this PR, I intend to work next on implementing dd (delete line) and, possibly S (substitute line) as natural extensions, so am looking forward to seeing this merged.