llvmPackages_{14,15,16,17}.lldb: use SWIG 4; swig3: drop #341384
Conversation
These have been failing to build on Hydra for over two months. It looks like it might be a Python 3.12‐related thing, but the SWIG 4 patches wouldn’t apply cleanly and these old versions should probably just be dropped anyway.
github-actions
bot
added
6.topic: python
6.topic: llvm/clang
The patches are already included in LLVM 16 and 17.
emilazy
force-pushed
the
push-rzyzyqyruqnt
branch
from
96ecbfe
to
fce8148
Compare
| @@ -1482,6 +1482,9 @@ mapAliases ({ | |||
| swift-im = throw "swift-im has been removed as it is unmaintained and depends on deprecated Python 2 / Qt WebKit"; # Added 2023-01-06 | |||
| swig1 = throw "swig1 has been removed as it is obsolete"; # Added 2024-08-23 | |||
| swig2 = throw "swig2 has been removed as it is obsolete"; # Added 2024-08-23 | |||
| swig3 = throw "swig3 has been removed as it is obsolete"; # Added 2024-09-12 | |||
| swig4 = swig; # Added 2024-09-12 | |||
There was a problem hiding this comment.
A bit curious about the rationale for making swig4 an alias rather than leaving it as a definition in all-packages.nix?
There was a problem hiding this comment.
My hope is that we can stop carrying multiple versions, so this is to forbid pinning the current version in‐tree, with a view towards dropping the alias. If you think that’s too ambitious then I can put it back (though the convention would be swig_4, which is another reason I felt like we should just drop it). So far it’s seemed like the number of things that break from new versions is relatively limited, and no one package has seemed like it has taken an inordinate amount of effort to fix during a normal staging cycle.
There was a problem hiding this comment.
I think it should be fine given recent experience? (Just wanted to confirm the intent was in fact to forbid pinning.)
There was a problem hiding this comment.
Yeah, my thinking is that we can always move it back if SWIG 5 somehow breaks everything, but that at present nobody really has any way of knowing that’s likely so pinning is a bad idea. (Same reason I unpinned all the FFmpeg 7 users.)
emilazy
force-pushed
the
push-rzyzyqyruqnt
branch
from
fce8148
to
932e637
Compare
|
Updated to include the LLDB patches for Swift too; building that locally on |
This version is no longer used for anything in the tree.
Let’s see if we can’t keep only one version of this going forward. `swigWithJava` has been an alias since 2009(!), so let’s just drop it.
This no longer uses the `swig3` package.
emilazy
force-pushed
the
push-rzyzyqyruqnt
branch
from
932e637
to
db5f130
Compare
|
I’m bored of waiting for Result of 7 packages marked as broken and skipped:
41 packages built:
|
|
(Swift finished building on |
|
nixpkgs-review happy, x86_64-linux: |
Description of changes
Turns out we do need the upstream patches after all, but it was relatively pain‐free. Hopefully we can drop LLVM 12 and 13 entirely soon.
Result of
nixpkgs-reviewrun on x86_64-linux 17 packages marked as broken and skipped:
20 packages built:
Things done
nix.conf? (See Nix manual)sandbox = relaxedsandbox = truenix-shell -p nixpkgs-review --run "nixpkgs-review rev HEAD". Note: all changes have to be committed, also see nixpkgs-review usage./result/bin/)Add a 👍 reaction to pull requests you find important.