In D24988#552276, @prj_rootwyrm.com wrote:

Also, there is a reason msbuild is to not be used for any reason: it not only does but will break. msbuild has never been tested beyond "yes, everything went terribly." It has never been in CI and will never be in CI, either. All CI and unit testing is done with monolite. msbuild is just a thing that exists.

Switch to ONLY_FOR_ARCHS as requested

Fix portlint issues for lang/mono

In D24988#551785, @mikael wrote:

Do the same for lang/mono please.

Fix portlint issues

Implement changes requested in review

Add msbuild patch

Update pkg-plist.powerpc as requested

By the way, this breaks the devel/msbuild port, because the old msbuild version is not compatible with the newer compiler version. Given that it relies on pulling a plethora of binary dependencies from nuget anyways, there is arguably little point in having it actually compile msbuild. On Arch Linux there is a package that simply pulls in http://download.mono-project.com/repo/ubuntu/pool/main/m/msbuild/msbuild_16.5+xamarinxplat.2020.02.20.11.54-0xamarin2+ubuntu2004b1_all.deb (ubuntu package provided by the mono project). This runs just fine on FreeBSD with only two changes:

Actually update to the corret btls patch

What is the status of this? Do you need me to split the changes? Would I need to create an entirely separate phabricator diff then?

Feel free to break the change into multiple patches. The change to use a TAILQ could certainly be committed on its own as a "no functional change" diff.

You're right, this should be a linked list. The only reason I went with the reallocation approach is that I wanted to touch as little code as possible to avoid new bugs. What I have now is a compromise: it leaves the old logic (allocate all list elements at creation, some of them are just invalid) untouched but it's still a list so append doesn't have to reallocate and fix pointers. I chose a TAILQ because the code assumes that the list is ordered so append has to insert at the tail.