Alex Rønne Petersen

> I'll publish a blog post on my experience soon™️ * https://www.alexrp.com/p/unpacking-and-repairing-the-tera-executable * https://www.alexrp.com/p/spring-cleaning-in-the-unpacked-tera-executable

Have you had a chance to consider this one further? I understand it's probably a fairly complex ask, and there's plenty of other things needing attention... just wondering how likely...

Upstream issue appears to have been fixed fwiw.

FWIW, as BN is gaining more and more sidebar widgets (especially now with Sidekick) and the right sidebar is actually fully usable, I think this issue is becoming more pressing....

Is there any reason in principle why it couldn't work? Obviously there would be some responsibility on my own part to root some `Type`s/`MethodInfo`s that I pass into DotNext.Metaprogramming, but...

Hmm, I don't quite follow. Wouldn't you just need to annotate the API surface of the library with `[DynamicallyAccessedMembers(...)]`, `[DynamicDependency(...)]`, etc as necessary? (And then plumb those through wherever the...

It seems there was some work done recently related to open generics: https://github.com/dotnet/runtime/pull/96648 That said, I don't actually know if this at all addresses your specific example of preserving `Task`...

> In this case all I have is to mark entire class with `RequiresUnreferencedCode` attribute I suppose applying dependency attributes to the `ForEachExpression` class itself would not work? > and...

Ok, I see what you mean. Thinking a bit more about it, I can't think of a good solution there either. What do you think about best-effort annotations? I.e.: *...

> I tried this approach several times and it requires a lot of work to analyze each case accurately. If I can find the time to go through the library...