@ringabout Even with new backports, the nightlies are still failing with:
os.nim(678, 30) Error: getApplOpenBsd() can raise an unlisted exception: ref OSError

Since this happens only on the nightlies, and not with the regular Nim CIs or to me locally, it is hard for me to reproduce.

Last successful nightlies build was before I started doing backports from this list, so I'm guessing the problematic commit is somewhere in the bottom 1/3 of this list (the oldest commit is at the bottom). Based on the commit messages in the list, can you maybe tell what might be causing it or, alternatively, how to fix the error?

Found it!!
I need to backport nim-lang/Nim#22585, which deals with exactly this error.

@arnetheduck The work on this list of commits has been completed. The "missing backports" (onec without a checkmark on the list) are commits which cause problems on the version-2-0 branch (each has been backported and then tested).

As it is more than a month since you made this list, if you want — make another list with recent devel commits which you would like to see in Nim 2.0.x, and I can work on backporting them.

@narimiran great, thanks! I'll take a look in a bit

@metagn nim-lang/Nim@9416595 - what would it take to backport this one?

@metagn nim-lang/Nim@9416595 - what would it take to backport this one?

Should be no trouble afaik, a direct cherrypick could work

Should be no trouble afaik, a direct cherrypick could work

Ok, let me try to backport it again. Maybe this time (with more commits backported) it'll work.

Should be no trouble afaik, a direct cherrypick could work

When I cherry pick your commit, tests/concepts/tmatrixconcept.nim fails with Error: unhandled exception: index 1 not in 0 .. 0 [IndexDefect].
That test file is the same in devel, so it's probably some other commit(s) that need to be backported for this to work.

Then I would say nim-lang/Nim@e8092a5 might fix it, but there may have been a backport attempt on this already?

Then I would say nim-lang/Nim@e8092a5 might fix it, but there may have been a backport attempt on this already?

I've tried it again now, on top of the previously mentioned commit, and it still doesn't fix the error in tmatrixconcept.nim.

Next idea? :)

None off the top of my head, I'll try to figure out what's failing from the stacktrace when I get the chance

The issue is with sempass2 not being able to handle the weirdness of the type of the type keyword, changing:

https://github.com/nim-lang/Nim/blob/fcb8461efab2ef7bdd976f82af8c7d1390f502ac/compiler/sempass2.nim#L891

to:

    if a.kind != nkSym or a.sym.magic notin {mNBindSym, mFinished, mExpandToAst, mQuoteAst, mType}:

(i.e. adding mType to the excluded sym magics) fixes it. My guess is the "type refactor" PRs dealt with this some other way but they are pretty big. We could push this change to just 2.0 or we can add it to devel then backport it for symmetry as it should be harmless.

I forgot to mention nim-lang/Nim@9416595 also has a followup nim-lang/Nim@4b1a841 and probably needs another one down the line to address nim-lang/Nim#23386.

I forgot to mention nim-lang/Nim@9416595 also has a followup nim-lang/Nim@4b1a841

Now I backported that followup added mType to the line you mention.

tests/concepts/tmatrixconcept.nim now passes on my machine, but let's see if CIs find any other problem.

let's see if CIs find any other problem.

No problem found. All mentioned commits are now available in version-2-0 branch.

With @metagn's fix (nim-lang/Nim#23571) backported, I consider this list of commits done.

@arnetheduck, please open a new issue with the recent commits you would like to see backported.