Conversation
Yep, I can confirm that this is unrelated to NUKE. Looks like .NET 10 has brought some massive degradation in terms of stack space handling. I guess this is related to stack-allocated reference types. Anyway, I don't think we can do anything but bite the bullet and lower the recursion depth. Based on my tests, 650 is still fine for release mode on Windows.
Which tests are those? I don't have any other failing tests in my local dev env. BTW, as for NUKE in general, do you know about this: nuke-build/nuke#1564 (comment) ?
This sounds pretty concerning. What do you think about the future of NUKE? In light of this announcement, is it worth staying on this train in your opinion? |
Hmm, interesting. That probably would explain this new behavior, even though surprising.
This was my bad, I have automatic EOL cleanup on in Rider and this caused the output to be different for the test case as returned value did not have trailing whitespace.
I'm aware of the situation, the latest 10.1.0 enabled most of the new NET 10 scenarios like slnx format etc so I'm not too concerned. NUKE is a mature project which allows you to work around things if needed. For me it's still the go-to choice until/if any blockers arise. |
|
Well, then I think we're good to go. Thanks for the update! |
The NET 10 tests are not passing for me locally of on GitHub Actions. The Release mode stack size test needs to go down to 395 like the Debug one. I think it's unrelated for this but very confusing. Also 4 parsing tests are failing after that. @adams85 are you experiencing any problems locally?
I earlier notified you on wrong PR, sorry for that.